----------------------------------------------------------------------- ; Copyright (C) 1995 ; Associated Universities, Inc. Washington DC, USA. ; ; This program is free software; you can redistribute it and/or ; modify it under the terms of the GNU General Public License as ; published by the Free Software Foundation; either version 2 of ; the License, or (at your option) any later version. ; ; This program is distributed in the hope that it will be useful, ; but WITHOUT ANY WARRANTY; without even the implied warranty of ; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the ; GNU General Public License for more details. ; ; You should have received a copy of the GNU General Public ; License along with this program; if not, write to the Free ; Software Foundation, Inc., 675 Massachusetts Ave, Cambridge, ; MA 02139, USA. ; ; Correspondence concerning AIPS should be addressed as follows: ; Internet email: aipsmail@nrao.edu. ; Postal address: AIPS Project Office ; National Radio Astronomy Observatory ; 520 Edgemont Road ; Charlottesville, VA 22903-2475 USA ----------------------------------------------------------------------- This file must be used to record any changes made to the code on the Charlottesville VAX in the logon area [AIPS]. The format to use is: line 1: Date Program name Your name line 2: What was done line 3: Where has this change been moved (e.g. 15OCT90) line 4: Blank The details of line 1 are important since programs are used to find specified strings and do conversions on them. It MUST be typed as follows: 1 blank, 4 digits, a period, two blanks, then the Month, day, comma, year, some space, a title, some more space, your name. Changes to 15APR91 *********************************************************************** Changes while 15APR91 was TST: *********************************************************************** *********************************************************************** 6305. September 24,1990 MK3IN Athol/Bill C Extensive improvements and enhancements. 1. Extensions for Spectral Line Data. Some changes in the Fourier transform conventions were neccessary in handling multiple correlator module lag functions. In particular the lag order convention needed to be reversed to concatenate the output of the modules. This produced several other changes in the Fourier transform of the lag functions to cross-power spectra which are discussed in a separate memo. A difference in sign convention between FOURG and QUIKFT was resolved. The cross-power spectra are reversed in frequency and have opposite sign of phase with respect to the original MK3IN. The frequency order is consistent with the known single dish spectra and the sign convention is consistent with the SAO program MK3MK2 (which claims to use the NRAO convention). AC functions are treated separately and are corrected for bias and clipping effects before being transformed to AC spectra. Auto-correlation functions are typically generated by cross-correlating half the XC lag range. Individual correlator modules can however be placed in auto-correlation mode and this may be used in the future. Some changes in the code were neccessary to deal with the two methods of generating AC functions. Correlation functions which do not have the same number of lags as the output file are skipped and a warning is printed. This is neccessary to avoid 8-lag fringe finding scans which are often included in the output tape. A warning message is also printed if the centre lag of the correlation function is non-zero. Magic value blanking is used in the lag domain to avoid transforming lag functions with missing lags. 2. Multiple Observing Bands. The new MK3IN produces an FQ table and allows multiple observing bands. 3. FBS Corrections Numerous test were carried out to determine the FBS model exactly. The new version should now do this correctly. 4. Checks on A-tape File Structure. Some checks were introduced to ensure that the A-tape file structure is as expected by MK3IN. The A-tape file headers are decoded and the MK3 file parameters are extracted. A warning is printed if the tape is a member of a multi-volume SAVEM set, as this may cause loss of data. The type 50 records within each extent are required to be in the expected order and only complete scan-baseline headers are used. These checks should diminish the effect of tape errors which may cause loss of synchronization and will also alert the user to any future changes in the MK3 data format. 5. Data Rejection Criteria The data rejection criteria used by FRNGE at the Haystack correlator were implemented in the new MK3IN version. A breakdown of the error statistics is available for each scan which can optionally be indexed on correlator module serial number. This feature was requested by the Bonn correlator group for spectral line data. 6. Baseline Indexing The MK3 tape format is baseline-based and consequently there is a high degree of redundancy in source- and antenna-based parameters which is the format that AIPS requires. The new version of MK3IN assumes that source and antenna parameters in files for different baselines may not be consistent and carries out several checks in this regard. The data are not rejected or changed in any way but a warning message is printed. The correlator model input parameters (or closely associated quantities) are also monitored throughout the run. A summary of the correlator model is printed at the end of the run. These consistency checks should help to isolate problems in the data set caused by changes in the correlation strategy or when merging data correlated at different times or different correlators. These checks may also alert the user to any unnanounced changes in the MK3 output tape format. 7. New Input Parameters Some minor changes in the input parameters were unavoidable but these were kept to a minimum. The APARM parameters are now: APARM(1) CL table increment (min). APARM(2) Maximum parity error rate (def: 0.01) (previously PE count). APARM(3) Maximum dropout rate (def 0.5). Correlator modules are rejected if the number of bits correlated is less than this fraction of the expected correlator count. APARM(4) ge. 1 => No FBS (previous APARM(3)) APARM(5) Max no of files on tape to process. (0 = until EOF). This is useful for short test runs where it is important that task terminates normally and writes the full complement of tables. APARM(9) Print level (0,1,2,3). See below. APARM(10) Debug flag (as before). Experiment file parameters: STATIONS Station names (as before). TIMEOFF IAT-UTC (as before). NO_LAGS Accept only lag functions of this length (if not present this will be set by the first lag function encountered). This parameter was introduced to avoid problems when the first scan is a fringe finding scan. NO_IF Number of AIPS IF's in output file. Introduced for similar reasons to NO_LAGS. Optional, if not given then this is determined from the first scan. SIDEBAND Force double sideband structure in AIPS output file (if this parameter is 'DOUBLE'). Optional, the sidebands will be concatenated if both USB and LSB data are present. STOKES from set {'RR','LL','RL','LR'}. Define the Stokes range of the output file. The largest contiguous range within the parameters is used to define the output file. Optional, if not given this will be determined from the first scan. FREQCODE MK3 character codes for {RR,LL,RL,LR} in that order. These codes are not very standardised. A warning is given if an unidentified MK3 code is encountered. Previously these data were assigned the lowest Stokes value. (def: 'R','L','r','l'). The optional parameters defining the AIPS output file structure were introduced to deal with tapes containing test scans as the first files (eg AC tests) and to force consistency. 8. Print Levels The print levels available are: 0: Print only a source header for each scan. 1: Include a one line error summary for the scan. 2: Print the MK3 file names and extent numbers. 3: Produce a error summary for each correlator module. 9. Miscellaneous The time indexing of each data record required modification due to a few minor typographical errors and rounding errors. It was also discovered that data records may not have time marks at integral correlator accumulation periods for the nominal scan start time. A scan start time offset is accumulated for each scan and the time mark is written for the AP midpoint. The HP number conversion routines were checked against known HP number reprresentations. A minor error in the conversion of single precision reals was found. A correlator index is generated from the 20XX module cross-reference table before processing each new type 51 data extent. This allows video converters to be multiply assigned in a given scan and should allow XC/AC mixing. The delay offsets can be in arbitrary order in the cross-ref table. 10. Code Consolidation The large common blocks were broken up into smaller units to reduce considerable compiler overheads (debug tables) and the preamble comments were updated and consolidated. Moved to 15JAN91 and 15OCT90 this date. 6306. September 24, 1990 VLA cont. calib. test Bill C. Added a new suite of POPS scripts to time and verify the correctness of the AIPS continuum calibration software especially as used for VLA data. These tests are patterened after the DDT tests and consist of two RUN files: VLACLOAD.001 and VLACEXEC.001; the first to read in procedures and the second to execute the tests. The two HELP files are VLAC.HLP and VLACSAVE.HLP. The test data and the master copy of the answers are on a FITS tape and were computed on a Sun 3/60. The test will run in 15JAN91 and later releases of AIPS. The tasks tested in some modes are: UVLOD, IMLOD, FITTP, UVCOP, TABED, UVFLG, SETJY, CALIB, TASRT, TASAV, GETJY, CLCAL, PCAL, TACOP, CLCOR, LISTR, SPLIT, UVDIF, HORUS and COMB. Moved to 15JAN91 this date. 6307. September 24, 1990 PCAL, PCAL.HLP Bill C. Added option (CPARM(1)) to average IFs. Moved nowhere. 6308. September 24, 1990 Bad characters in POPS Bill C. Added traps for bad characters in INPUT, GO, TPUT, TGET, SHOW and TELL by adding calls to VERSTR in AU1A, AU2, AU2A. Moved nowhere. 6309. September 24, 1990 Bad character traps Bill C. Added two new routines for trapping bad characters: APLSUB:CHKCHR.FOR Checks that all characters in a string are one of the following: A-Z, a-z, blank, 0-9 and +-*?_=!@#$%^&(){}[]:";~`,./<>| AIPSUB:VERSTR.FOR Checks that a character string or array in POPS memory contains only valid characters. Moved nowhere 6310. September 24, 1990 AVSPC Phil After much negotiation with Neil we have finally settled on what AVSPC should do with the reference frequency, it leaves it alone, so the reference pixel is updated to reflect freq. changes due to channel selection. Also the bandwidth in the output file is the input bandwidth * the number of channels averaged. Moved to 15OCT90 and 15JAN91 today. 6311. September 24, 1990 BPASS Phil There were two problems remaining with LL polarized line data: (1) If the user had a dual polzn database and requested solutions for LL alone, an indexing error caused the channels to be shifted to the right by one, this meant that the last channel in the internal buffer was occupied by the u, v, w of the next record, this caused large closure error messages to be printed. (2) Also in a dual polzn database, when dividing by an external channel 0, another indexing error caused the LL line data to be divided by the RR continuum. I discovered that I had also broken the ability to generate seperate BPASS entries for multiple bandpass calibrator scans. Moved to 15OCT90 and 15JAN91 today. 6312. September 24, 1990 ACFIT Phil I added the ability to use the SNVER adverb, by incorporating a call to the CALSEL subroutine. In the course of debugging this I fixed a couple of other problems: (1) the last entry for a particular run of ACFIT had the incorrect source number; (2) if the user just wanted to solve for a single antenna rather than all antennas, there were logic problems, although it still worked; (3) if there was no valid data at all for a particular solution the program printed confusing error messages, but again still worked. Moved to 15JAN91 this date. 6313. September 24, 1990 New adverbs Phil Two new task-specific adverbs, XAXIS is used in SNPLT to specifiy which type of axis is to be plotted on the abscissa; DOERBAR is used by VBPLT and SNPLT to specify the use of error bars on the plots. Modified POPSDAT.HLP and NEWPARMS.001. Moved to 15JAN91 this date. 6314. September 24, 1990 AIPSPUBL:COOKZ.TEX Bill J. Changes at the AOC necessitated changes in this document. (1) OS 8.1 tape handling is somewhat different. Changed this for nrao1 also. (2) AIPS logical disks were renamed a couple of weeks back. (3) An additional disk was added on yucca. (4) Printer queues were changed to allow use of QMS PS 1500 laser printer. (5) Various typos and errors in layout were fixed. Moved nowhere. 6315. September 24, 1990 UVBAS Juan Uson/Bill C Fixed a problem with flagged channels. If a channel was flagged, no higher channel numbers were processed in averaging the "continuum" channels. Moved to 15JAN91 this date. 6316. September 24, 1990 UVPLT Phil Added the user option (BPARM(10)) to include the autocorrelation spectra in any plot made. The default is not to use that data. Modified the help file accordingly. Moved nowhere. 6317. September 24, 1990 MAKMAP/UVPREP/HORUS Glen MAKMAP was modified to call UVGRTB which will grid any size problem (up to 4096). Unfortuately an assumption in a subroutine UVPREP was not correct for HORUS (although it was correct for MX.) UVPREP fixed and HORUS relinked. Moved to 15JAN91 this date. 6318. September 24, 1990 MX Glen Surprise. A long standing bug was uncovered. MX did not scale U,V,W to second IF for data selection. This caused some data to be outside ALGSTB, ALGSUB limits in gridded subtraction. Particularily frequent when working with multi-array data. Modified only MX subroutine MXSEL. Moved to 15JAN91 this date. 6319. September 25, 1990 AVSPC Phil A philosophical argument with Neil was finally resolved (in Neil's favour unfortunately), and the calculation of the new reference pixel was changed. Moved to 15OCT90 and 15JAN91 today. 6320. September 26, 1990 AIPSHELP.HLP Bill C. Next Designated AIP cycle. Moved nowhere. 6321. September 26, 1990 MK3IN.FOR Athol Extended visibility buffer size from 1024 to 2048. Also some changes to cope with null A-tape records (type 50, id=0). Moved nowhere. 6322. September 27, 1990 MK3IN.FOR Athol Correct bug in data rejection criteria. Moved nowhere. 6323. September 30, 1990 APGNOT:FILLM.FOR Gareth This new version of FILLM represents several months' reworking and debugging. The major changes are as follows: 1. The keywords associated with the multi-source files have been changed from LINECHAN and CORRMODE to OBSCHANS, CORRMODE, SELCHANS, and VLAIFS. This is to permit more sophisticated selection criteria for appending data to existing datasets. OBSCHANS number of channels on VLA archive tape CORRMODE the VLA correlator mode SELCHANS channels selected from OBSCHANS VLAIFS the names of the VLA IFs used (OBSCHANS and SELCHANS are not used for continuum data.) A multi-source file is checked for all of these parameters as well as the Band, # polarizations, 1st Stokes' parameter, and # IFs to determine whether data should be appended to the file by FILLM. This change will mean that this version of FILLM cannot append data to databases created by previous versions. 2. The naming convention for output files has been simplified. The classes are '? BAND' for continuum, and 'LINE' and 'CH 0' for spectral line. Classes such as 'CH 0 1' are obsolete. 3. A new source (NX) entry is now forced whenever the Start LST of the observation changes. This is how the VLA on-line system marks a new scan. This forces a new source in some cases where the older versions did not. It has a two main advantages: it reliably detects a source scan change without checking for the almost infinite possible changes in the set-up of the observations; and it allows QUACK to be run subsequently for each change of the on-line system configuration. The only disadvantage, that logically contiguous data are now in separate scans, can be easily addressed by running INDXR post facto. 4. The subroutine MCWANT, which controls the selection of data and the changes in configuration, has been completely recast. Selection criteria that are always in effect are timerange (and date), program name, and VLA subarray. Selection criteria that may be overridden by DOALL=1 (true) are observing mode, number of channels per baseline, frequency, and band. 5. All possible spectral line observations with the VLA are now fully supported. The flexibility now built in to the VLA on-line system is reflected in the complexity of the subroutine MCINI. There are 26 different ways (3 continuum and 23 spectral line) of writing the data into up to 8 multi-source files in parallel. (An additional 8 ways, for mixed continuum and spectral line data, are supported here but not yet provided by the VLA). 6. The derivation of the number of channels per IF has been moved from subroutine MCINI to MCH1, where most other control information from the VLA archive tape is decoded. Bugs fixed include: 1. FILLM did not always correctly respond to a configuration change, such as FQ number or correlator mode, when averaging data. This was helped by the simpler and more rigorous scan change criterion (see above). A change introduced in February 1990 in the 15APR90 version had this unreliability as an undesirable side effect. 2. A baseline record that was split between to physical tape records had a jump in phase. The subroutine MCGBRC miscounted by 1 the number of data points (real and imaginary being counted as 2) left at the end of the first record. This change was also introduced in February 1990 and was not noticed for several months except by two careful observers. It did not affect continuum data at all. Since the VLA has been in the larger (A and B) configurations since that time, there has not been much use of the spectral line modes. Moved to 15JAN91 this date. 6324. October 1, 1990 APLSUB:LMPIX.FOR Bill J. IMLOD failed when trying to read an IMPS(!) tape due to divide-by-zero in LMPIX when setting up the reference pixel. Trap added. Moved nowhere. 6325. October 3, 1990 New directories/logicals Bill C. Added several new directories and corresponding logicals in AREAS.DAT. ...APL/DEV/UNIX/BERK/IBM (APLIBM) IBM RS/6000 Z routines ...APL/DEV/UNIX/BERK/DEC (APLDEC) Decstation Z routines ...Q/DEV/PSAP/IBM (QIBM) IBM RS/6000 Q routines ...Q/DEV/PSAP/DEC (QDEC) Decstation Q routines ...INC/NOTST/IBM (INCIBM) IBM RS/6000 includes ...INC/NOTST/DEC (INCDEC) Decstation includes ...SYSTEM/UNIX/IBM (SYSIBM) IBM RS/6000 shell scripts etc. ...SYSTEM/UNIX/DEC (SYSDEC) Decstation shell scripts etc. ...APL/CONTRIB (APLCONTR) is a directory for contributed software that is not to be included in the standard installation but is available for users to compile, link and use. Source and HELP files go in this directory. Moved to 15JAN91 this date. 6326. October 4, 1990 POSSM Phil Minor logic error caused the freq. labelling of the x-axis to fail. Fixed. Moved to 15JAN91 this date. 6327. October 9, 1990 Contributed code from Leiden Walter Jaffee/Bill C A number of tasks and their help files were put in directory APLCONTR where they can be manually compiled and linked. These tasks may not run (or even compile) on a given machine. These are: WSLOD - Read Westerbork data tapes RDCAL - Westerbork style redundancy calibration. SAD - Finds sources and fits Gaussians to an image. CCNTR - Plots symbols at positions of components in a CC table the way CNTR plots ST tables. STARC - Read a CC table from an ASCII table (the analog of STARS for ST tables). Moved to 15JAN91 this date. 6328. October 11, 1990 POSSM Eric Changed the setting of plot scales to avoid the 360-degree phase ambiguity when possible and to give some slight borders around the data. Also corrected an error message. Used the latest version from CV to return to CV, but did not keep it here. It claims to do phase shifting for position offsets but does not support sprectral-line data - and this in a spectral-line program! Moved from Oz this date to 15APR91 and 15JAN91. 6329. October 11, 1990 SPFLG, TVFLG, Calibration Eric Changed: SOUFIL It did not close the SU table on error and did not examine the kind of error from ISTAB but simply assumed that any ISTAB error meant non-existant SU table (a non-error condition). Now it will close the table and treats most ISTAB errors as real errors. OTBSRT Changed it to write a scratch table as the output and then to rename it to the desired output name only if everything has worked. Otherwise, when a file was sorted onto itself it could be destroyed by any error. TABSRT Removed close of input table - it is always needed now to go to scratch table. Corrected a recent addition to the code --- it is okay to exit quickly when there are no records, but only if the OUTVER is the same as the INVER. SPFLG Changed to call SOUFIL only when it is relevant and to honor its error return. Before it called it always and ignored the error return. TVFLG Changed to call SOUFIL only when it is relevant and to honor its error return. Before it called it always and ignored the error return. Moved from Oz this date to 15APR91 and 15JAN91. 6330. October 11, 1990 SPFLG, TVFLG Eric Changed them so that when they apply the flags to the main FG table they remove them from the master grid file. This means changing the FC table to zero entries and going through the master grid file and magic-value blanking all SPFLG'd or TVFLG'd pixels. Moved from Oz this date to 15JAN91 and 15APR91. 6331. October 11, 1990 LISTR Eric Changed SCAN display to show number of visibilities for each scan in the NX table list whenever the display is wider than 79 characters. Moved from Oz this date to 15JAN91 and 15APR91. 6332. October 11, 1990 SETJY Eric Cleaned up the typing and display of the input values in the history routine. Added 1934-634 for AT folks and increased the formula to 4th order. Did other clean ups of typing and logic, including getting all calculated fluxes into the HI. Moved from Oz this date to 15APR91 only. 6333. October 11, 1990 SPFLG, TVFLG, TVFIDL Eric Changed SPFLG to display time and channel in all flagging modes. Otherwise, the flux displayed doesn't mean much. Also changed SPFLG to force TVPSEUDO to display some color initially. And changed TVFIDL to do have better initial values and hence not zoom further when it starts zoomed and to do black and white first even if a zoom was done ahead of all coloring. Changed TVFLG as SPFLG. Moved from Oz this date to 15APR91 and 15JUL91. 6334. October 11, 1990 SCHOLD Eric Corrected SCHOLD to move 4 characters to upper case so that "quit" would be accepted as such rather than as an (illegal) AIPS verb. Moved from Oz this date to 15APR91 and 15JAN91. 6335. October 11, 1990 AVSPC Bill C. Added option to average in IF as well as frequency by calling routine AVGIF. Added adverbs AVOPTION, BIF and EIF. Modified the description of the function of the task in AVSPC.HLP. Moved nowhere. 6336. October 11, 1990 New adverb AVOPTION Bill C. Added new adverb AVOPTION (averaging option) in POPSDAT.HLP, NEWPARMS.001 and AVOPTION.HLP. Moved nowhere. 6337. October 11, 1990 AVGIF Bill C. New routine in APLSUB to average a visibility record in IF and frequency, selecting frequency channels and a range of IFs. Moved nowhere. 6338. October 11, 1990 UVAVG Bill C. Modified not to take weighted sum of source and FQ id parameters. This putting garbage into these variables. Moved to 15JAN91 this date. 6339. October 11, 1990 APLIBM:* SYSIBM:* INCIBM:* QIBM:* Brian Checked Kerry's IBM RS/6000 modifications into the system. The files were: APLIBM: D2IBM.FOR ZCREA2.C ZGNAME.FOR ZTAP2.C ZTPWAD.C D2IEEE.FOR ZDATE.C ZLPCL2.C ZTIME.C ZTQSP2.C S2IBM.FOR ZDCHI2.FOR ZR32RL.FOR ZTKILL.C ZTTOPN.FOR S2IEEE.FOR ZDCHIC.C ZR64RL.FOR ZTPCLD.C ZTTYIO.FOR ZABOR2.C ZDEAXF.FOR ZRLR32.FOR ZTPMID.C ZTXMA2.C ZABORS.C ZFRE2.C ZRLR64.FOR ZTPMIO.FOR ZGNAM2.C ZTACT2.C ZTPOPD.C (ZACTV9 is checked out by Chris - I'll put it in later). INCIBM: DAPC.INC DBUF.INC DSEL.INC QIBM: QCFFT.FOR QINIT.FOR QRLSE.FOR QCLNSU.FOR QMULCL.FOR WHNALT.FOR SYSIBM: AIPS ASSNLOCAL.SH FCOPTS.SH LIBR.DAT ZLPCL2 AIPXAS CCOPTS.SH INCS.SH LINK ZTRLOG.C ASOPTS.SH CDVER.CSH LDOPTS.SH OPT2.LIS ASSNLOCAL.CSH CDVER.SH LIBR PP Moved to 15JAN91 this date. 6340. October 15, 1990 DOCTXT:UGUIDE,VGUIDE.RNO Bill C. Updated for 15OCT90 release. Moved nowhere. 6341. October 16, 1990 IBLED Phil Several new features and some bug fixes: (1) IBLED now has a mode in which it can create catalogued uv work files to which you can return for an editing session. (I call this its 'fire-and-forget' mode). (2) It now creates all its necessary scratch files at the start of the session to avoid disc space problems later. (3) The antennas, baselines adverbs are now used to select antennas to be edited. (4) The handling of the FC tables has been changed, previously when an FC table existed reading it and applying its contents was a painfully slow business. This table is now sorted to time order and the application is much more efficient. (5) IBLED could not apply the flagging information to single source compressed files, fixed this. (6) When averaging in time there was an error which caused one more visibility to be averaged each time, i.e. you asked for 2 minutes averaging time, you got three. (7) The times in the FG table were equal to time +/- interval/2, this was wrong it is now time +/- interval. (8) A display buffer was not being cleared so that when a new baseline or frame was displayed with a smaller number of points the old data from the previous buffer load was still visible. Also changed IBLED.HLP Moved to 15JAN91 and 15OCT90 this date. 6342. October 17, 1990 CVEL Phil I ran a profile on cvel and discovered that 28% of the cpu time was spent in TABIO, this was due to the fact that the CL table was read from the beginning for each visibility until the doppler freq. offset for a particular time was found. I changed this to keep an index of where I was in the CL table for each antenna and CL time and interval, this reduces the amount of TABIOing needed dramtically. Moved to 15JAN91 this date. 6343. October 17, 1990 MX Glen Slight modification to correct Max, Min UVrange to allow second IF to be included in UV data if first IF fits in allowed range. 6344. October 17, 1990 ALGSTB/UVGRTB/UVGRID/UVAPAS Glen Corrections to routines to correctly handle multi-if data for case when not all map fits in "AP". Essentially requires multi-channel passes for two IFs widely separated. Single IF and case of all map in "AP" are not affected by the changes. Will move to 15JAN91 in a week. 6345. October 17, 1990 DDTLOAD.001 Glen Added declaration of adverb TERSE to user environment. If TERSE greater than 1, then PRTAC is NOT run. Moved to 15OCT90 and 15JAN91. 6346. October 17, 1990 IMLOD/UVLOD/FITTP Brian To work around a bug in the Sun f77 compiler, the number -2147483648 was changed to -2147483647 - 1. (In 2s complement most_positive_int < abs(most_negative int)). Moved to 15JAN91 and 15OCT91 (tape) this date. 6347. October 19, 1990 XVSS Chris XVSS may now optionally use the MIT shared memory extension to the X Windows System. This should speed up XVSS on machines with large amounts of physical memory. In order to use the shared memory extension you must have a version of Unix that supports System V shared memory: SunOS supports this as an option (the kernel may have to be rebuilt on some systems). You must build XVSS with SHMOPT set to -DUSE_SHM in the Makefile and add the line: {\tt xvss.useSharedMemory: True} to the X resources database before starting XVSS (the X resources database is normally initialised from {\tt .Xdefaults}). If this line is missing or {\tt xvss.useSharedMemory} is set to {\tt False} shared memory is disabled. This is recommended for machines with smaller memories since the use of large amounts of shared memory may cause severe paging problems. Also replaced the Makefile, which was missing from the .SHR file and made the following minor changes. - Excessive blank space has been removed from the control panel. - XVSS will now read the colour of the graphics plane from the {\tt xvss.graphicsColor} (for Americans who can't spell) resource in the X resource database. The value of {\tt xvss.graphicsColor} may be a named colour from the X colour database or a hexadecimal specification in the normal X11 #RRGGBB format. The default colour is now green (the previous purple colour has a poor contrast against the blue dominated AIPS spectral colourmaps, particularly on slides); users with the old style Sun greyscale monitors that run off the red gun may wish to set {\tt graphicsColour} to {\tt red} or {\tt white}. - XVSS now has OPEN LOOK help available for all components. This is limited at the moment since the current version of XView does not support scrolling help windows but it does provide an on-line reminder of the keyboard accelerator bindings. The help text is provided in a seperate uuencoded file, xvss.uu, because the bloody checkin system on CVAX won't accept the long lines of text. Moved to 15JAN91 this date (because of missing Makefile). 6348. October 19, 1990 APGNOT:FILLM.FOR Gareth A bug in routine FLMDAT was causing the CH0 and LINE datasets to differ in the number of visibility data points. This annoyed BPASS considerably. The bug only occurred when the last visibility of a given time-stamp was flagged bad. The logic error had also caused the first good visibility after a flagged visibility in a given time-stamp to be rejected! This had not been noticed. The bug only affected spectral line data, but the missing visibility occurred in continuum data as well. Moved to 15JAN91 and 15APR91 this date. 6349. October 22, 1990 APGNOT:DAYFX.FOR Glen Array crash caused times to be set to zero. Previous version of DAYFX fixed the DAY but left time 0, so TVFLG and many other tasks would not work. DAYFX now fixes day as before, but if, after fix, the time is still less than previous, the time is set to previous time. No data flagging is done. Moved nowhere. 6350. October 22, 1990 ZLOCK in APLBERK Chris Added a conditional compilation branch that uses the POSIX fcntl interface to file locking if _POSIX_SOURCE is defined. As POSIX advisory locks use a separate mechanism from the BSD flock() interface this may clear up the problems due to apparently spurious locks under Convex OS. Unfortunately, the separate mechanism also means that there is a risk of files being corrupted or information being lost until all AIPS tasks have been linked with the new variant. In case the locking problems do not vanish the POSIX code branch will attempt to get information on any conflicting lock and print the ID of the process that has locked the file; this is not robust --- it is possible that the locking process could release a lock in the interval between ZLOCK detecting that the file is locked and ZLOCK requesting information about the lock (the PID will then be meaningless). Moved nowhere pending thorough testing. 6351. October 22, 1990 APGNOT:FILLM Gareth Fixed two tests in the selection criteria that were reversed (qualifier and channel code) and led to the exclusion of the requested data rather than the inclusion. Moved to 15JAN91 and 15APR91 this date. 6352. October 23, 1990 POSSM Phil When the phase centre shifting option was invoked the shift done was a constant across the band. This was in error the u, v, w's used to calculate the shift should have been scaled with frequency for each channel and the phase rotation calculated with the scaled values. This is now done, the error in most cases was fairly minor since the frequency difference across the band is normally a very small fraction of the observed frequency. Another minor bug, when the user wished to plot the BP table and specified a source name then that was compared with the sourcename in the table, if they didn't match the data were not selected for plotting. I had missed the case of the source number in the table being 0, in which case it should be plotted. Fixed that. Moved nowhere. 6353. October 25, 1990 APLSUB:CATKEY Brian There was an off by one error so that an illegal type (of 0) was being accepted. Moved 15JAN91 this date. 6354. October 25, 1990 APLSUB:SCREAT Brian SCREAT was modified to put the SYSTEM name (SYSNAM in DDCH) in any scratch files it creates in 3 keywords, SYSNAME1, SYSNAME2 and SYSNAME3. The idea is that we only want to automatically destroy scratch files created on the current cpu since we can't yet interrogate processes on remote machines. Moved nowhere. 6355. October 25, 1990 AIPSUB:DESCR Brian Check the system name keywords in the scratch files and only delete files that either belong to this CPU or don't have the keywords in the scratch file (so this system and the old one can coexist without the disk filling up with scratch files). Moved nowhere. 6356. October 25, 1990 APLBERK:ZTQSP2 Brian Use a better heuristic to cut down on the number of aips only processes (something at the end of a line that has capital letters and ends in a number). Moved to 15JAN91 this date. 6357. October 26, 1990 SYSIBM APLIBM INCIBM Brian SYSIBM:AIPS - source rather than execute AIPXAS SYSIBM:AIPXAS - allow XAS display on remote hosts. APLIBM:ZACTV9 - ZACTV9 was out and couldn't be checked in earlier. APLIBM:ZFRE2 - AIXized so that free now works. APLIBM:ZTQSP2 - Spy now works APLIBM:ZMOUN2 - Mount issues a rewind and selects the correct device for 1600 or 6250 density. In general this will need to be changed from site to site, althout most machines won't have a 1/2" tape drive. INCIBM:DSEL.INC - No longer needed by xlf. INCIBM:DBUF.INC - No longer needed by xlf. Move to 15APR91 this date. 6358. October 29, 1990 APLIBM:ZLPCL2.C Brian Removed this stubbed version so we can print from the ibm. Removed from 15JAN91 this date. 6359. October 30, 1990 APLNOT:SOUELV.FOR Glen Limited hour angle output range to between -12 and 12 hours (-pi to pi). Moved to 15JAN91 this date. 6360. October 31, 1990 INC/DAPC.INC QPSAP Glen Modified the QPSAP Pseudo-AP memory size include file, Increasing the default size to a total of 256 K Words. The main AP size is 64 K and the secondary is 192 K. The subroutine QPSAP/QINIT was modified to automatically take advantage of the larger size. Moved Nowhere. 6361. October 31, 1990 QPSAP/QCLNSU.FOR Glen Modified this routine to eliminate conversions from integer to real in most cases. Significantly speeds machines with fast floating point operations. If savings is significant, this will be moved to new. Moved Nowhere. 6362. October 31, 1990 DOCTXT:USERNO.LIS Gareth The file USERNO.LIS is used by various site-specific routines to display the local system use. It contains the NRAO user number list. This has not heretofore been part of the standard distribution for reasons that I do not understand. It is now. Moved to 15JAN91 this date 6363. October 31, 1990 AIPS' K array Eric Changed POPSGN, AIPS, AIPSB, and AIPSC to use a 950-word array in the K array for temporary litorals. The previous limit of 250 was too little for Bill's new DDT clone. Changed SGLOCA to force a new SG-file version. Moved from Oz this date to 15APR91 only. 6364. November 1, 1990 DOCTXT:APPENDIXC.TEX Bill C Updated description of thecontents of the GRfile to include the one-line summary and e-Mail entries. Moved nowhere. 6365. November 1, 1990 CLBSNR Phil In the self-cal mode of CALIB, and when no closure errors are requested to be printed, CALIB changes the no closure errors to huge closure errors just to catch ridiculous values. This is OK but the user then gets a message on his/her screen for every integration. Changed this to print out the closure error message header only when there is an error. Moved to 15JAN91 this date. 6366. November 1, 1990 APLNOT:NCALC Phil NCALC assumed that routine DNRM2 was still a function but it was changed to a subroutine months ago. I'm surprised no one tripped over this before, I suppose it shows how often the GCON option in CALIB is used. Moved to 15JAN91 this date. 6367. November 2, 1990 SPFLG Eric Corrected error which caused an incorrect upper channel number to be used for actual flagging when in the one-channel mode. Moved from Oz to 15JAN91 and 15APR91. 6368. November 2, 1990 SGLOCA Eric Corrected bad code for POPS error handling. Moved from Oz to 15JAN91 and 15APR91. 6369. November 2, 1990 AIPS Eric Corrected to actually abort on an error in INIT and not to say "ABORT!!!" when the Sg LASTEXIT file is too old. Moved from Oz to 15JAN91 and 15APR91. 6370. November 2, 1990 SQASH Phil Apparently SQASH has never worked when a sub-image in the first two coordinates was requested. The code looped from 1 -> NY (where NY = # output pixels in y direction) whereas it should have looped from BLC(2) to TRC(2). The loop in the x-direction was OK. Moved to 15JAN91 today. 6371. November 2, 1990 APLPGM:UVDIF.FOR Glen Has been broken for some time for UN-Compressed UVdata. (At least uncompressed UVdata in the DDTs.) Recent fix to allow Compressed data caused it to die for Uncompressed, so Compressed data checks are skipped for uncompressed data. Moved to 15JAN91 this date. 6372. November 2, 1990 APGNOT:MK3IN.FOR Athol K. Updated the maximum number of correlator module entries expected in the correlator module cross-reference table for each baseline. The MK3 documentation suggested that that 100 was a reasonable choice but a tape was encountered with 112 entries (new maximum = 200). Moved nowhere. 6373. November 2, 1990 AIPSPUBL:COOK11.TEX Phil Added a section on fringe-fitting MkIII data. Also changed references to CALIB to FRING where relevant. Moved nowhere. 6374. November 6, 1990 UVLOD Phil Athol and I encountered a situation where the array decleration for VISPRT in subr. UVFDAT was not big enough. Changed it to be 4096 rather than 512. Moved nowhere. 6375. November 6, 1990 FITTP Bill J. Subroutine EXTWRT was not writing flagged table rows to tape (IERR = -1 from TABIO in "READ"). Consequently, there was a mismatch between the number of records written in "NAXIS2" and what was actually on the tape. This caused PRTTP problems (and UVLOD too, I would think). Now, FITTP writes ALL rows to the output buffer. Also corrected a typo: "ISORTORD =" --> "ISORTORD=". Moved to 15JAN91 today. 6374. November 7, 1990 FITTP Bill J. The act of writing all rows to the tape effectively unflags the flagged rows (see 6375) so instead FITTP now finds the actual number of unflagged rows and writes these only to tape. The value of NAXI2 is the number of unflagged records --- i.e. the actual number transferred to tape. Obviously, the fact that the contents of the tape are different from the disk file means that this is at variance with our usual philosophy. This is temporary and should be revised when the FITS tape writer is modified for VLBA data. Moved to 15JAN91 today. 6375. November 7, 1990 CLBSNR Phil Moved the printing of the closure phase error header message down 2 lines, in its current position it printed the header > 1 times before printing the closuer errors. Moved to 15JAN91 this date. 6376. November 7, 1990 FILLM Bill J. Added call to CHLTOU (case conversion) for the observing program id prior to comparison with the user-supplied character string. This needed for the rare cases where the program code is lower case due to bug in OBSERVE. Moved nowhere. 6377. November 8, 1990 CONVL Bill C. Fixed several logic errors that kept OPCODE='DGAU' from working. Also the values of T and F (true and false) were not set in CONDIV. CONVL probably still doesn't work in this mode but at least it runs to completion. Moved nowhere. 6378. November 8, 1990 APLNOT:TABFQ, TABNDX Gareth Minor changes in the declarations of EQUIVALENCEd arrays in TABFQ and TABNDX so that they can be used, in debugging, with array bounds checking turned on. Moved to 15APR91 and 15JAN91 this date. 6379. November 9, 1990 APLIBM:ZDELA2 Tom Pauls/Brian Tom Pauls has supplied a nice ZDELA2 (based on sleep/usleep) that actually works. I put in a paranoid error check. Moved from NRL to 15APR91 and 15JAN91 this date. 6380. November 10, 1990 IBLED Phil More new features and bug fixes. (1) Users had complained about the way IBLED switched between the editing and menu screens, this caused a lot of flashing on the IIS and IVAS and was painfully slow on the Sun's. Changed so that now the menu's appear on the same screen as the editing frame, they are always visible. Makes the process much more efficient. (2) Added a new menu option: SELECT FRAME - which enables the user to select his/her own frame to be edited by positioning the cursor on the total vis. fn at the top and define a start and stop position., this is then displayed in the main editing frame. (3) Added an on-line help facility, when users point to a menu item and hit button D a help message is printed on the terminal. (4) The user can now specify the OUTDISK to which the catalogued uv workfile goes, 0 => lets IBLED choose. (5) Added some checks to prevent users hitting the wrong buttons in some of the editing modes. (6) Discovered that when a user came to re-edit a file using an already existing catalogued file his/her previous editing appeared not to have been applied. Actually the editing info was present in the FG table but when reading a catalogued uv-file the FG table is not applied. So when finishing a session the FC table is copied to the uv work file before being translated to an FG table and deleted from the master file, so when a new session is started the FC table is copied back to the master file and applied before any data is displayed. Also updated the helpfile. Will move to 15JAN91 soon. 6381. November 11, 1990 VBPLT, VBPLT.HLP Phil Converted VBPLT to work with multi-source compressed data if so desired. It can now apply all the calibration adverbs on the fly in the same manner that other routines can. One limitation that had to be built in is that when plotting a model against the data I force the user to specify a source name if the data is a multi-source file. Also modified the helpfile to relect the changes. Moved nowhere. 6382. November 13, 1990 IBLED.HLP Phil Put back the wrong version of IBLED.HLP on the 11th, the number of adverbs in the helpfile was not the same as the task expected. Moved nowhere. 6383. November 13, 1990 TVFLG.FOR TVFLG.HLP DTVF.INC Brian Modified TVFLG to write/read its grid file on the disk pointed at by IN2DISK. Moved to 15JAN91 this date. 6384. November 13, 1990 LISTR Bill J. Another LL only bug. GAIUV1 and GAINUV subroutines get the indexing of the columns wrong in the solution table if there is one only polarisation in the database and it is 'LL'. Moved to 15JAN91 today. 6385. November 14, 1990 APGNOT:FILLM Gareth Two major changes: 1. if the user is silly enough to try to force FILLM to append to a file which has incompatible data, FILLM now stops; 2. all character fields read from the data tape are forced to upper case. Also fixed a pedantic complaint: DOCONCAT is now true only if >0. Moved to 15JAN91 this date. 6386. November 14, 1990 MAPIT.001/MAPIT.HLP Glen Added set of AIPS procedures called MAPIT in the $RUNSYS area. HELP MAPIT explains the inputs and the procedures are run by typing MAPIT. First round of procedures, extensive changes antisipated. Moved Nowhere. 6387. November 14, 1990 APGNOT:UVPRM/CCFND Glen Two New Tasks. UVPRM has three functions, to find a reference antenna, the maximum UV-spacing in the data and to find the UV-spacing at which the average flux of the source is greater than the input FLUX. CCFND finds the last clean component before a negative clean component then finds the clean component FACTOR brighter than the negative component. These tasks are useful for setting self-calibration parameters. Moved Nowhere. 6388. November 15, 1990 APLSUB:IMERGE Brian LISTR was not properly selecting baselines because of a bug in IMERGE.FOR - unique antennas were being rejected as duplicates. Besides being buggy, the code was quadratic although this wasn't necessary. Moved to 15JAN91 this date. 6389. November 15, 1990 APGNOT:IMMOD Brian The conversion of degrees to radians for BPA was done in exactly the wrong sense. Moved to 15JAN91 this date. 6390. November 15, 1990 IBLED Phil When flagging an area and hitting button B to flag another area a logical was not being reset so that you had to reselect the menu option. IBLED should also now work under SSS and XAS. In the helpfile changed OUTDISK -> IN2DISK to match what Brian did to TVFLG. Moved nowhere 6391. November 15, 1990 CLBSNR Phil Finally fixed up CLBSNR to write the closure error message only onece each time the routine is called. Moved to 15JAN91 today. 6392. November 16, 1990 CVEL Phil The AC functions were being shifted in the opposite direction to the XC functions, this didn't matter too much for VLBA stations but did for intercontinental experiments. Moved to 15JAN91 today. 6393. November 19, 1990 APLNOT:FRQTAB Brian Fix up FRQTAB so that it will work correctly if the IF and Frequency axes are in non-standard order. (Relinked FRING and HORUS to try to force problems if any to the fore). Moved nowhere. 6394. November 19, 1990 RUNSYS:MAPIT.001 Glen Replaced ASCAL and GNPLT with CALIB and SNPLT. Moved nowhere. 6395. November 20, 1990 Corrections to Delay residuals Bill C. There were several places where corrections were being made to delay residuals without a corresponding correction to the phase residual. Since, in the presence of delay errors, the phase is defined as the phase correction at the reference frequency. When a delay residual is corrected the phase must also be corrected. This change was made in APLNOT:SNSMO.FOR, APGNOT:CLSMO.FOR and APLNOT:CALREF.FOR. The change to CALREF required a change to it's call sequence. This was required further changes in APLNOT:CALADJ.FOR, QPGNOT:CALIB.FOR and QPGNOT:FRING.FOR. Also added FQ id code to the call sequence to SNSMO. Moved to 15JAN91 this date. 6396. November 20, 1990 APGNOT:CLCAL.FOR Bill C. Changed call sequence to SNSMO. Also changed to allow OPCODE='SMOO' for multisource data. Moved to 15JAN91 this date. 6397. November 20, 1990 APLNOT:SN2CL.FOR Bill C. Removed the use of the weights in the SN table in determining the interpolation in the SN table. This practice could cause serious problems if the SN table weights (SNRs) were not in fact related to the data quality. An example of this is when, in one scan an antenna is used as the reference antenna in fringe fitting and thus has a low "SNR" = min acceptable + 1.0 and in the next scan was not the reference antenna and has a much higher weight="SNR". Moved to 15JAN91 this date. 6398. November 20, 1990 APGNOT:DECOR.FOR Bill C. New task to compute the decorrelation amoung a set of selected correlations. The output values have a real part that is the ratio of the scalar amplitude average of the selected correlations to the vector average. This progam is especially useful for looking for coherence problems in multi IF and/or multi channel data. This task allows application of calibration and/or editing and will write a new multi source file. Moved to 15JAN91 this date. Also DECOR.HLP 6399. November 20, 1990 APGNOT:SNCOR.FOR Bill C. New task to apply corrections to SN tables. The corrent version only contains the option to force the residual fringe rates to the average of the selected IF/polarization. This option appears necessary to maintain coherence between the polarizations for MkIII VLBA data. Moved to 15JAN91 this date. 6400. November 20, 1990 APLPGM:DIFRL.FOR Bill C. new task to divide the RR correlations by the LL correlations. This task is useful for looking for coherence and amplitude calibration problems between RR and LL correlations. It should be generalized to allow referencing phases to an arbitrary correlation. Moved to 15JAN91 this date. 6401. November 20, 1990 QPGNOT: FRING.FOR Bill C. Fixed number of words of pops adverbs to ask GTPARM for (it was about 30 too few). Also fixed indexing and numerous other problem that occured when averaging frequencies in the output of a single source compressed data file. Moved to 15JAN91 this date. 6402. November 20, 1990 APGNOT:SHOUV.FOR Bill C. New task. A clone of LISTR that does column listings of closure phases for selected triplets or listings with different IF/channels in different columns. These displays are very useful for detecting and analysing problems in multi IF/channel data sets. Also SHOUV.HLP. Moved to 15 JAN91 this date. 6403. November 20, 1990 APGNOT:GLENS.FOR Glen Created a Gravitational LENS modeling program. Makes models of a Blandford elliptical galaxy or a point mass (black hole). Also $HLPFIL:GLENS.HLP. Moved nowhere 6404. November 20, 1990 RUNSYS:DDTLOAD.FOR Glen Made the TERSE option alittle Terser. Does not print error messages separately if TERSE is true. Unchanged if TERSE<=0. Moved nowhere 6405. November 21, 1990 TVFIDL Eric Correct to use the correct zoom factor when restarting in an already zoomed mode. Moved from Oz to TST only. 6406. November 21, 1990 APGNOT:FETCH FETCH.HLP Frank Ghigo/Brian A bug was fixed that only allowed square images to be input. The number of characters allowed in a line was bumped up to 16k to allow larger images to be Fetched. Moved to 15APR91 and 15JAN91 20-Nov-1990. 6407. November 21, 1990 APLNOT:FQFTAB Bill C./Brian This returns the FQ ID/IF table of frequency offsets and channel increments. This is Bill's original version and hasn't been really tested yet. Moved nowhere. 6408. November 21, 1990 FLAGGING LL DATA Phil Found and fixed a bug that caused FG tables attached to LL only data not to be applied. In DATFLG the loop that applied the flag ran from 1 -> NCOR, where NCOR is 1, but the logical array inside the loop was set up as 0,1,1,1 i.e. to flag all L correlations, so it only read the 0 and jumped out of the loop. Changed UVGET to record the ICOR0 of the input data and added a pointer inside the DATFLG loop which depends on ABS(KCOR0) - 1. This meant a change in DSEL.INC with the agony that accompanies that. Also found that BPASS and SDTUV used a variable called KCOR0, changed its name. Moved nowhere. 6409. November 26, 1990 APGNOT:DECOR Bill J. The count of NPARMS was low by 1 so OUTDISK never functioned. Also fixed a couple of typos in the HELP file. Moved to 15JAN91 today. 6410. November 27, 1990 Gripe verb and file Bill C The gripe verb now asks for an e-mail address and a one line summary of the gripe. These are the 8th and 9th entries in the GR file. Changed AUC.FOR and CHGRIP.FOR in AIPSUB. Moved nowhere. 6411. November 27, 1990 UVDIF.FOR Bill C. The data being printed was read from the wrong buffer for compressed data causing very erratic results. Added a second decompression buffer and now data to be printed are always copied to and used from the decompression buffers. Moved to 15JAN91 this date. 6412. November 27, 1990 APLUNIX:ZBYTF2 Brian ZBYTF2 was determining if the input and output arrays were the same incorrectly. Also, the case of no byteflip was not being handled properly (but ZBYTFL short-circuits this anyway). 6413. November 27, 1990 RUNSYS:MAPIT.001 Glen Added a few more adverb initializations and comments. Moved nowhere. 6414. November 27, 1990 APLSUB:FNDXY.FOR Glen Removed a subroutine from UVPRM.FOR (identical to one in UVPLT.FOR). Modernized and commented. Moved nowhere. 6415. November 27, 1990 APGNOT:UVPRM.FOR Glen Changed binning of UV-data for calculating source flux and added more comments. Removed subroutine FNDXY.FOR Moved nowhere. 6416. November 27, 1990 APLGEN:ZR32RL Brian ZR32RL was doing byteflips but not word flips which caused the Decstation to turn perfectly good IEEE numbers into garbage. The other 32 and 64 bit real conversion routines will need to be similarly changed (32 bit ints are wordflipped!). Moved to 15JAN91 this date. 6417. November 28, 1990 FRING Bill C. Removed unused adverb GAINERR. Also FRING.HLP Moved nowhere. 6418. November 28, 1990 MX.HLP Bill C. Improved documentation about sort order. Moved nowhere. 6419. November 28, 1990 SPLIT.HLP Bill C. Improved documentation of APARM(2) (the input data averaging time). Moved nowhere. 6420. November 28, 1990 HORUS Robert Braun/Bill C. HOURS was only applying uniform weighting corrections to the first of each group of images. Move to 15JAN91 this date. 6421. November 19, 1990 ZR32RL ZRLR32 ZR64RL ZRLR64 Brian None of these worked for the decstation because although the bytes were properly flipped, the words weren't. Moreover the blank and zero checks were wrong as well. Moved to 15JAN91 this date. 6422. November 30, 1990 FILLM.FOR, FILLM.HLP Gareth 1. Fixed bug (denkfehler) introduced when the multi-colored VLA spectral line options were added. Two VLA IF spectra from different polarizations (e.g. A & D) were put into the same AIPS IF even if the frequencies were different. They are now correctly separated into different multi-source files. 2. Allow a timerange as short as 10 seconds for debugging purposes. 3. Added code to deal with VLA holography measurements. Moved to 15APR91 and 15JAN91 this date. 6423. December 1, 1990 QPSAP:QMULCL.FOR Glen Ed Fomalont found that under rare conditions one of several fields being mapped with MX would have NO residuals loaded into the AP. When this happened, QMULCL, would not correctly locate the max for the next clean cycle. Added logic for fields with NO map pixels in AP. Would not effect the convex which use a different QMULCL. Will move to 15JAN91 at the end of the week. 6424. December 1, 1990 QIBM,QSUN:QCLNSU.FOR Glen First Sun specific Q routine, which does minor cycle beam subtraction. Should be 20% faster than PSAP routine for SUN's and the IBM's which are strangely slow at real to integer conversions. For large DDT test, MX is in QCLNSU 30% of time. Moved nowhere. 6425. December 3, 1990 APLGEN:ZRHPRL.FOR Athol Small error in the conversion of HP reals (a part in 10**6 or less). The mantissa is a 2's complement signed integer and should be decremented by 1 if the sign bit is set. Moved nowhere. 6426. December 3, 1990 QPSAP:QMULCL.FOR Glen Error in location of first maximum in minor cycle loop. Moved nowhere. 6427. December 3, 1990 RUNSYS:DDTLOAD.001 Glen Yet another adjustment in TERSE so that some results are always printed. Moved to 15JAN91 this date. 6428. December 3, 1990 ANCAL Bill C. Implemented the advertised capability to give baseline dependent calibration factors by writing a BL table if "BASELINE" cards are present in the calibration file. This implementation allows only one set of baseline factors per uv data file. Also improved the documentation in ANCAL.HLP. Added adverb BLVER to specify the output BL table. Fix an indexing error for LCP only data when POLNO=2 in the input file. Moved nowhere. 6429. December 3, 1990 AU7 (RENAME) Bill C. In it's check for duplicate file names RENAME was including type; thus it was possible to have an MA and a UV file with the same name, class, sequence and disk number. This causes problem for a number of verbs and tasks (e.g. IMHEAD) which no longer have INTYPE as an adverb and always pick the one it finds first in the catalog. Moved nowhere. 6430. December 4, 1990 GETJY Bill C. Added adverbs SOUCODE to allow specifying a source "cal code" and SNVER to specify a particular SN table. Did some general cleanup of the code. Moved nowhere. 6431. December 4, 1990 ZABOR2.C Brian I changed every ZABOR2.C, i.e. in APLUNIX, APLBERK, APLBELL, APLSUN, APLCRI, APLIBM. Instead of handing the signal handler a function we have been typically giving it an uninitialized external function pointer which through a quirk of the linker has normally been resolved correctly - everywhere but on the IBM and Decstation. I am checking everything into TST and the IBM version only into New. Moved nowhere (mostly). 6432. December 4, 1990 BLCAL Bill C. Added a trap in BLCDIV to barf and die if multiple sources are selected and/or a flux density to be used is missing. Moved nowhere. 6433. December 4, 1990 APLSUB:LPOPEN Brian A harmless looking unneeded common caused the Decstation to be unhappy so I removed it. Moved Nowhere 6434. December4, 1990 APLDEC INCDEC QDEC SYSDEC Brian The routines to make the Decstation work under AIPS. QDEC is currently empty, i.e. pure QPSAP. APLDEC: (Notable: byte problem in ZXASX2, Dec doesn't use the unix convention for recordlength in ZTPOPD.FOR) ZABOR2.C ZCREA2.C ZDCHIC.C ZTPWAD.FOR ZXASX2.C ZABORS.C ZDCHI2.FOR ZTPOPD.FOR ZTTOPN.FOR INCDEC: (1MW=4MB PSAP) DAPC.INC SYSDEC: (The usual) AIPXAS CCOPTS.SH INCS.SH QMSLAND.HDR ZLPCL2.QMS ASSNLOCAL.CSH CDVER.CSH LDOPTS.SH QMSPORT.HDR ASSNLOCAL.SH FCOPTS.SH LIBR.DAT ZLPCL2 Moved to 15APR91 and 15JAN91 this date. 6435. December 5, 1990 VBPLT Bill C. Improved default values for TIMERANG and UVRANGE which weren't very good. Also removed the correction in component positions since this correction is already done by the CLEANing tasks. Moved nowhere 6436. December 5, 1990 FNDPOL Bill C. Improved documentation and added a test to see if the data is a single source file and if so it assumes that polarization calibration has already been done. This should suppress many of the bad warning messages. Moved nowhere 6437. December 5, 1990 CGASET Bill C. Now sets LCALTM to the curent time if only doing BL calibration; this avoids many unnecessary calls to CGASET. Also fixed logic for calling BLSET. These bugs were causing time variable BL calibration to fail. Moved nowhere. 6438. December 5, 1990 BLSET Bill C. Now resets IBLP1 if not applying BL cal. This aviods many unnecessary calls to BLSET. Moved nowhere. 6439. December 5, 1990 BLCAL Bill C. BLCAL could write many blanked solutions if using an averging time less that the whole data set. These blanked solutions corresponded to times when there was no data. Now, if a gap is found no invalid solutions are written for it. Moved nowhere 6440. December 5, 1990 LISTR Bill C. In 'LIST' mode LISTR now assumes that data with no AN table comes from the VLA and has antenna numbers up to 28. Also forced new pages between amplitude and phase GAIN listings (DPARM(1)=5) and between sets of baselines in 'LIST' mode. Changed the default averaging time for MATX listings of single source files to be 10 min. rather that the whole data set. Also change the description in LISTR.HLP Moved nowhere. 6441. December 5, 1990 PRTTP Bill C. Fixed heuristic test for VLA Modcomp archive tapes to account to translations of old format tapes; format revision numbers between 3 and 50 are now accepted. Also fixed extension file routine to know about 'A3DTABLE' and '3DTABLE' types. Moved nowhere. 6442. December 6, 1990 CLCAL,SN2CL Bill C. Added option INTERPOL='SELF' for "self" calibration. In this mode there is no smoothing of the SN table and only solutions from the same source is used to calibrate CL table entries. If the two SN table entries surrounding each CL table entry are both from the same source then the closest one is used. This is implemented in APLNOT:SN2CL.FOR. Also changed CLCAL.HLP. Moved nowhere. 6443. December 7, 1990 APLNOT:GCALC1.FOR Bill C. A typo in the phase only solution portion caused the soluton to become unstable and sometimes give bad solutions. It's pretty amazing that it ever worked at all. This was the L1 solution routine for CALIB. Moved to 15JAN91 this date. 6444. December 7, 1990 QPGNOT:CALIB.FOR Bill C. In it's computation of the mean gain modulus for the gain stabilization CALIB was only checking that the weight of a solution was positive. Routine CLSNOU was changing the weights of flagged reference antenna data to positive values so the blanking value was added to the sum. This caused a very large mean gain modulus and thus very small amplitudes of data corrected using this mean gain modulus. Also added messages about which solution type. Moved to 15JAN91 this date. 6445. December 7, 1990 DOC:VMSNOTES.90D Bill C. Added information about TVFLM and IMFLM problems and the need to use SETPAR to set the character size for TK devices. Moved nowhere. 6446. December 7, 1990 AN, SU tables Phil There was a problem which had bitten some users caused by the introduction of FQ identifiers. It was possible for users to a) determine fluxes of sources for one freqid, write them in the SU table and apply them to data with another freqid, without being aware of it; b) determine antenna polarziation characteristics with PCAL for one freqid and, again, apply them to data with another freqid. The solution to this problem was to add another keyword to the SU and AN tables. When FREQID dependent values are changed in these tables a FREQID keyword is written with the value of the FQID for which the values were changed. When applying these values to data this keyword is checked against the FREQID of the data itself and will not allow the user to apply the values if they are not the same keyword. If the values have not been changed the FREQID keyword in the tables will have the value -1. This change neccesitated the following changes: (1) Includes DANT.INC, DSOU.INC, DANS.INC had a new variable added to define the FQID in the SU and AN tables. (2) SOUINI and ANTINI were changed to read/write the new keywords and to ensure that old data with no keyword in the tables could still be read with no problem. (3) POLSET was changed so that it would check the compatibility of the table FREQID's with the data before applying the antenna polzn corrections. (4) DGHEAD, FNDSOU, GETSOU, SOUFIL, SOUFND, SOURNU, SUSEL, GETANT, GETNAN, ANTDAT, DATPOL were all subroutines which called either ANTINI or SOUINI and were changed to deal with the new call sequences. (5) SETJY was modified to write a keyword value in the SU tables if necessary. (6) PCAL was modified to write keyword values in both the SU and AN tables if necessary and also to check the compatibility of application of any previous corrections. (7) CLCOR, DBCON and GETJY were modified to check the compatibility of FREQID keywords for their applications. (8) The following tasks called SOUINI or ANTINI and had to have their call sequences changed: AVER, AVSPC, BLCAL, BLFIT, BPASS, CALIB, CVEL, FILLM, FILLR, FITTP, FRING, GAPLT, HORUS, IBLED, IM2UV, LISTR, MK3IN, MULTI, POSSM, PRTAN, PRTGA, PRTUV, SDTUV, SETAN, SPLIT, TVFLG, UVAVG, UVCOP, UVFIL, UVFIX, UVLOD, UVSIM, VBMRG, VBPLT, VLBIN, WSLOD. (9) All tasks that load data from an external format were changed to write the default value of the keyword in the SU and/or AN tables. Moved nowhere, Eric to change TVFLG. 6447. December 7, 1990 APGNOT:MK3IN.FOR Athol MK3IN now deals with the case where the correlation functions are determined over a lag range that is not symmetric about zero lag. The maximum symmetric lag range is extracted in this case. Also updated for new keyword FREQID in AN and SU tables. 6448. December 7, 1990 IPC Routines Chris The following routines form the minimal set of interprocess communications routines for AIPS. The current routines only support TCP stream sockets under BSD Unix. In APLGEN: - ZROPEN --- open a connection to a server - ZRCLOS --- close a connection - ZRREAD --- read a message - ZRWRIT --- write a message - ZROPN2 --- low level open (stubbed) - ZRCLS2 --- low level close (stubbed) - ZRRD2 --- low level read (stubbed) - ZRWR2 --- low level write (stubbed) In APLBERK: - ZROPN2 --- open a TCP socket - ZRCLS2 --- close a socket - ZRRD2 --- read a message from a socket - ZRWR2 --- write a message to a socket In APLCVEX: - ZROPN2 --- open a TCP socket (requires special version) Moved nowhere. 6449. December 7, 1990 DBCON Athol There existed a problem with VLBI data that the AN tables in the two files may be inconsistent, other checks (listed below) were also implemented. 1) The AN tables are merged if they have compatible AN table headers and the data are to be written as one subarray. The antennas are not renumbered although a warning will be given if their numbering is incompatible. Antennas which appear in both tables are checked for consistency. Entries with coordinates equal to DBLANK or with a station vector of < 2m length are assumed to be null. These changes are contained in a new subroutine DBANT (some AN table code was moved from HISANT to DBANT). If the AN tables are not compatible then the AN tables are dealt with as before. The algorithm for separate subarrays should be unchanged. 2) In compiling the source number translation table I also included a check on the consistency of the coordinates of sources found in both input tables. A warning is printed if they disagree but no other action is taken. 3) For data containing multiple FQ ID's we need to check that the FQ tables were in agreement. I added a subroutine DBIF to give a warning if any existing input FQ tables disagree (does not deal with CH tables). There is one difference between this version and the existing DBCON. At present DBCON will force the AN header ref date to agree with the catalog ref date, irrespective of the subarray parameter epoch. This version writes the input AN header that is closest to the cat ref date. Moved nowhere. 6450. December 7, 1990 UVAVG Athol Updated accumulation buffer indexing to handle files containing both XC and AC data. Data for different FQ ID's are now accumulated separately. This is useful when averaging AC function over long periods in a data file with FQ switching. This will work with TB or BT sorted data as before. The limitations on averaging are: TB data: NANT * (NANT+1) * NFQID * (NIF * NCHAN * NSTOKES * 3) / 2 <~ 10**5. BT data: NFQID * (NIF * NCHAN * NSTOKES * 3) / 2 <~ 10**5 Moved nowhere. 6451. December 8, 1990 IBLED Phil IBLED, like other tasks, was not correctly dealing with LL only data when reading the FC table. It now does. ALso when a user asked for all data to one antenna to be flagged that information was written correctly into the FC/FG table but internally the program was unaware of it and so it appeared as if the editing had not been done when a subsequent baseline was displayed. Moved to 15JAN91 along with the whole TST version. 6452. December 10, 1990 DBCON Phil The previous changes to DBCON (entry 6449) were a little too severe in that warning messages were printed out if there was the slightest discrepency in source or antenna positions between the files. The first user to try this had antenna positions differing by a few mm and was confused as to the messages. Changed it so that the warnings are only printed if source positions differ by > 10mas, or antenna positions differ by > 1cm. Also changed the MXANT parameter from 27 to 45 to fix another minor bug. Moved nowhere. 6453. December 10, 1990 AIPSPUBL:COOKZ.TEX Brian Describe the IBM RS/6000 (Lemur) that now sits in CV. 6454. December 10, 1990 HELPFIL:ISCOMP.HLP Bill J. List of tasks that can handle compressed (= packed) data. In response to Gripe 4492 and local whinges. Moved nowhere. 6455. December 10, 1990 HLPFIL:UVAVG.HLP Bill J. Help file was rather uncertain whether 'TB' and 'BT' data could be handled. In response to Gripe 4531. Moved 15JAN91 today. 6456. December 10, 1990 Expedited data Chris Added new routines to handle expedited data (also known as out of band data. In APLGEN: ZRXDWR.FOR -- write expedited data ZRXDW2.FOR -- low-level write (stubbed) In APLBERK: ZRXDW2.C -- low-level write (socket) Moved nowhere. 6457. December 10, 1990 ZLOCK in APLBERK Chris Removed debugging statements. Moved to 15JAN91 this date. 6458. December 11, 1990 COOK11, COOK6 -.TEXs Bill J. Added information on existence and use of IBLED. Removed references to CALIB for global fringe-fitting and inserted FRING instead. Moved nowhere. 6459. December 11, 1990 $YPGNOT/RMCUB.FOR + HLP Bill J. Added Greg Taylor's Faraday rotation curve fitting and plotting program to the system. Moved nowhere. 6460. December 12, 1990 $APGNOT/DBCON.FOR Athol Small change to ensure that AN tables with incompatible FREQID's are not merged. Moved nowhere. 6461. December 12, 1990 CALIB Bill J. Add solution summary to HI file. In answer to Gripe 4573. Moved nowhere. 6462. December 12, 1990 UV2TB Bill C. New task. Converts autocorrelation spectra to a table of type AS (Autocorrelation Spectra). Full calibration system included. This allows POPS procedures to read and write autocorrelation spectra. Moved nowhere. 6463. December 14, 1990 COOK4.TEX Bill J. Added 'P' and 'X' band calibration information. Moved nowhere. 6464. December 14, 1990 FILLM Gareth 1. Changed the method of determining whether antennas are present or not. The previous method was not valid for copies of old (pre-1988) tapes. 2. Two fixes to correct bugs in FQ table handling. One was a "mad bomber" - four arrays whose dimensions were too small. The second was the use of control information from VLA IFs A and B for IFs C and D. This had been necessary in the past. Since C and D are now, in general, independent of A and B, this assumption is not longer valid. It was causing incorrect frequencies, bandwidths, channel separations, etc. for experiments where independent tuning is used. 3. The velocity and reference frequency are now written into the SU table for channel 0 as well as the line data. 4. Changed FLMBC, which determines the VLA band code, to use the most up-to-date frequency limits. Other code to determine the band code has been replaced by a call to this subroutine. 5. Fix to ignore shadowing for raster scan (holography) data. 6. Problems with the VLA multiple-IF spectral line modes. The previous fix (November 30, 1990) went too far - now corrected. Fixes made in 15JAN91 and 15APR91 yesterday 6465. December 14, 1990 ANTINI, DANT.INC Phil Under certain subtle conditions an exisiting AN table could be opened for WRITE and all the header keywords could be zeroed. This would happen when the ANFQID on entry was set to zero, normally an illegal value for it but one I didn't trap. It now checks for this. Also changed the declerations of various arrays to use the MAXANC parameter in PUVD.INC, changed DANT.INC in same way. Moved nowhere. 6466. December 14, 1990 APLNOT:SOUINI.FOR, INC:PUVD.INC Bill J. Some arrays in SOUINI.FOR were explicitly dimensioned as 19 rather than MAXSUC. MAXSUC in PUVD.INC had been changed to 21 in the course of Phil's table revisions. This led to failures on spectra --- AOC Sparcstation 1 only. Changed SOUINI to use PUVD.INC and changed MAXSUC in PUVD.INC from 21 to 19. Revised in TST only since AIPS on most other systems seems to work and quarterly update takes place next week. Moved nowhere. 6467. December 17, 1990 INC:PUVD.INC Bill C. Changed MAXCHA (number of channels) to 2050. Moved to 15JAN91. 6468. December 17, 1990 CVEL Phil Mark Claussen discovered that CVEL did not work for data whose velocity type was heliocentric. That is not strictly true, the first data point was fine, after which the logical specifying if the data were heliocentric was reset to .false.. Fixed this. Moved to 15JAN91 today. *********************************************************************** Changes while 15APR91 was TST: *********************************************************************** *********************************************************************** 6469. January 13, 1991 APLNOT:DATGET.FOR Brian A "double equivalence" of the form EQIVALENCE(RECORD,RECI,RECI) caused the 3090 compiler to go into an infinite loop. Moved to 15APR91 this date. 6470. January 14, 1991 FITTP Brian The IBM 3090 complained about having -2**31 in Data statements. Changed to an assignment statement of the form: MAGIC4 = -2147483647 - 1. Moved to 15APR91 this date. 6471. January 14, 1991 AIPSUB:CHGRIP.FOR Bill C. Modified to blank fill the output buffer after writing. This keeps from writing junk in the last record of the file that was confusing the gripes processing software. Moved from 15JUL91 this date. 6472. January 14, 1990 APGNOT:ANCAL.FOR Bill C. Fixed logic error which caused LCP corrections for dual polarization data to be applied to the RCP portion of the CL table. Moved from 15JUL91 this date. 6473. January 14, 1991 TABINI Gareth The volume ID was not updated on write. This led to a fatal error when appending to a table on a disk whose number was different from the one on which it was created. This was only a problem when a disk was writable from two (or more) CPUs and had different volume IDs on the two (or more) systems. Moved to 15APR91 and 15JUL91 this date 6474. January 14, 1991 FILLM Gareth The decoding of the antenna status information was incorrect. The "reference antenna" status for raster scans (holography) was also incorrect. Both fixed. Moved to 15APR91 and 15JUL91 this date 6475. January 14, 1991 PRTTP Gareth Changes to the printing procedure for VLA archive tapes to handle pre-1988 data correctly. The "time at end of integration" is now used for all tapes instead of "time of geometry calculation," since the latter was does not exist on the old tapes. Moved to 15APR91 and 15JUL91 this date 6476. January 14, 1991 UVSEN Gareth The first call to OUTPRT had too many arguments. Moved to 15APR91 and 15JUL91 this date 6477. January 14, 1991 USERNO.LIS Gareth The latest list of AIPS users. Moved to 15APR91 and 15JUL91 this date 6478. January 17, 1991 TVFLM, IMFLM Gareth TVFLM and IMFLM are causing sufficient problems with their non-standard nature that I have removed them from YPGNOT and their corresponding HLP files from HLPFIL. I have put them into the contributed code area. Moved to 15APR91 and 15JUL91 this date 6479. January 17, 1991 RMCUB Gareth RMCUB has been removed from the system entirely because it contains proprietary code. 6480. January 17, 1991 SYSAIPS:AREAS.DAT Brian YXAS should be under YDEV so that the search scripts find it in the right place. Also removed an error in the contrib area for Gareth. Manually moved YXAS to the right place in the RCS directory, NRAO1 and lemur by hand. Other sites that use the midnight job should follow suit. Moved from 15JUL91 this date. 6481. January 22, 1991 APGNOT:SETJY Bill J. Formatting error bollixed output to terminal of calculated fluxes of primary calibrators. Moved from 15JUL91 today. 6482. January 23, 1991 QPGNOT/FRING Bill J. Inappropriate call arguments for subroutine DNRM2 caused FRING to bomb when PRTLEV=2 was selected. Moved from 15JUL91 today. 6483. January 24, 1991 CVEL Phil Minor modification to deal with MkIII VLBI data. The reference and remote station are defined in the opposite sense to MkII data. Moved nowhere. 6484. January 25, 1991 SYSUNIX:AREAS.*SH Brian AREAS.DAT had changed, but not the corresponding *SH files. Moved from 15JUL91 this date. 6485. January 28, 1991 FITLD Phil If the user asked for more files to be loaded than actually existed the task did not recognize the end of tape and died deleting the last file written. Fixed. Moved from 15JUL91 today. 6486. January 28, 1991 ANTINI Phil It appears that AIPS antenna tables thought they were ANNENNA tables, and have done so for a long time. Fixed spelling. Moved from 15JUL91 today. 6496. January 28, 1991 SYSIBM:AIPS,aips,AIPXAS Brian Changed AIPS script to not source AIPXAS when the user has asked for a remote TV. Also modified AIPXAS to start up XAS as a grandchild to see if this will fix up the strange AIPS hanging until the TV is killed problem. SYSDEC:AIPS was modified similarly. Moved from 15JUL91 this date. 6488. January 28, 1991 TVFLG Phil The new call sequence for ANTINI had not yet propoagted to TVFLG. Moved from 15JUL91 this date. 6489. January 30, 1991 FITTP.FOR FITTP.HLP Brian Changed FITTP to default to IEEE format, since for most machines that will be the fastest format, and also with scaled integers you can have problems if wildly wrong data is also in the dataset. Also changed the documentation. Moved from 15JUL91. 6490. January 30, 1991 SYSSUN/LIBR.DAT Pat Changed $YSSS to $YSS; this was screwing up the Sun Installation and causing lots of confusion. Moved to 15APR91 only. 6491. February 4, 1991 FUDGE Eric Added an example to the DIDDLE subroutine and a comment about it in the help file. Users need the guidance of code that at least copies VIS(*) to RESULT(*). The also makes FUDGE do something which can be tested via UVDIF. Moved from Oz to 15JUL91 and 15APR91. 6492. February 4, 1991 QHEADER Eric Changed KWIKHD to compute correctly offset coordinates to pixel 1 for UV data. It was displaying the reference value and claiming pixel 1 before. Moved from Oz to 15JUL91 and 15APR91. 6493. February 4, 1991 MATHS Eric For some bizarre and undocumented reason this task went to a lot of trouble to (incorrectly) restrict itself to one output plane. When fed a cube it died because of treating BLC and TRC as arrays of size 2. Fixed it to handle cubes correctly. Moved from Oz to 15JUL91 and 15APR91. 6494. February 4, 1991 UVLOD, R3DTAB Eric Changed all buffer sizes, doubling them. Also changed the pointer array in UVFDAT. It was dimensioned 512 and was supposed to hold pointers to all polarizations, IFs, channels, etc.; of course, it failed miserably. Also changed a data array in R3DTAB from 512 to very big. It was supposed to hold the largest table row in FITS form prior to conversion. The output array was big, not the input! Moved from Oz to 15JUL91 and 15APR91. 6495. February 4, 1991 POSSM Eric Corrected a bunch of stuff in the Fortran and help files: (1) Cleaned up the handling of logical input parms and their descriptions in the help file. (2) Changed fixed-scaling to allow self-scaling on one of the parameters. (3) Corrected the frequency dependent phase shift for SHIFT. Whoever coded this mess does not understand how the coordinates work. Instead of 1 + CATR(KRCIC+JLOCF) * (IC-1 + BCHAN - CATR(KRCRP+JLOCF)) / CATD(KDCRV+JLOCF) this person coded 1 + CATR(KRCIC+JLOCF) * (CATR(KRCRP+JLOCF)*IC - 1) / CATD(KDCRV+JLOCF) which has no relationship to reality. (4) Corrected a parameter which was set only in an uncommon IF clause, but then always used to zero a summing array. (5) Changed it to plot multiple IFs from its basic buffers rather than requiring multiple buffers of length MAXCHA * MAXIF. (6) Changed it to plot vertical lines between IFs in the multi-IF mode. (7) Corrected bug which caused plots of the BP table to stop at the first blanked channel. (8) Changed to omit spectra from the plot if the channel 0 value is less than BPARM(4) when dividing by channel 0. Cleaned up the messages some. (9) Added option to control the plot type - amp and phase, real and imag, or any one of the above. Changed help and AU8A accordingly. Moved from Oz to 15JUL91 and 15APR91. 6496. February 4, 1991 EXTLIST Eric Brought the EXTLIST code (in AU8A) up to date for POSSM and added some displays of SHIFT, CODETYPE (new), and the divide by zero option. Moved from Oz to 15JUL91 and 15APR91. 6497. February 4, 1991 AVER Eric Raised the buffer sizes to handle multi-channel data. Moved from Oz to 15JUL91 and 15APR91. 6498. 12 December, 1990 TVFLG, SPFLG Eric (1) Corrected one misleading format in SPFLG and one bad format in both SPFLG and TVFLG in the flag by form area. (2) Added IN2DISK to SPFLG, changing DSPF.INC, SPFLG.HLP, SPFLG. Picked up that addition from CV to DTVF.INC and TVFLG, and put a correct explanation of its effect in TVFLG's help file. (3) Changed display of status lines at the bottom of the menu to write all possible characters in each line. Otherwise, non-blank characters changed to blank did not get erased. (4) Doubled buffer sizes in TVFLG, SPFLG, GRIDTC, GRIDTB, TVFOAD, SPFOAD,a nd GTBWRT. Added a test to make sure that the gridding buffers are big enough in GRIDTB. They could handle only 4 subarrays before. (5) Also added remarks to TVFLG help file recommending using one subarray at a time and corrected the SPFLG help file to document its limitation to one subarray at a time. (6) Also increased number of baselines from 2000 to 5000 in TVFLG and DTVF.INC. (7) Changed SPFLG to correct interactive windows to all channels only when the set window is very close to the edge. Changed TVFLG similarly. (8) Corrected y position of CURVALU-like display in some modes in both tasks. (9) Corrected both tasks to flag all channels/baselines in TIME and TIMERANG modes even if only a subimage is currently (8) Corrected y position of CURVALU-like display in some modes in both tasks. (9) Corrected both tasks to flag all channels/baselines in TIME and TIMERANG modes even if only a subimage is currently loaded. Previously, it flagged only within the subimage in the work file but flagged the full range in the UV data set. (10) Corrected both tasks to handle compressed data in single-source UV data files. Before they would simply have messed them up. (11) Corrected GTBWRT, TVFLG, SPFLG, SPFOAD, TVFOAD to set the source number of empty rows to -1, rather than zero. This allowed a correction to the averaging for TV display so that averages only go over one source and the requested time interval. Single source files would average n times without regard for how much time passed between them. Moved from Oz to 15JUL91 and 15APR91. 6500. February 5, 1991 VLBIN Phil Modified to add data selection parameters based on bandwidth and observing frequency. Also changed VLBIN.HLP. Moved from 15JUL91 today. 6501. February 5, 1991 BPASS Phil Neil found a bug I had introduced in the recent overhaul of BPASS. A GO TO had been commented out. While checked out I added a mode to force the normalization of a bandpass function when requested, this is very useful for VLBI data where normalization by channel 0 is often not possible. Also updated BPASS.HLP. Moved from 15JUL91 today. 6502. February 7, 1991 AIPPGM:AIPMAN.FOR,AIPLAS.FOR Glen Removing AIPMAN.FOR and AIPLAS.FOR from $AIPPGM. These programs no longer work, but are placed in $AIPSCONTR Already done in 15JUL91. 6503. February 7, 1991 AIPPGM to AIPNOT Glen Moving all programs which do NOT use the TV in AIPPGM to AIPNOT. Currently these programs are linked several times and placed in the $LOAD/ALT* directories. Programs are AIPSC, AJAX, BATER1, DELSG, FIXCAT, FIXUSR, GR2TEX, GRIPR, GRITP, POPSGN, GMNGR, RDFITS, RECAT, SETPAR, SHOPA, SHOPB, and UPDAT. Already done in 15JUL91. 6504. February 7, 1991 $RUNSYS/MAPIT.001 Glen Added numerous comments and error checks. Moved from 15JUL91 this date. 6505. February 8, 1991 TVFLG Eric Corrected TVFOAD. In fixing the handling of sources and averaging, I left out an initialization statement, causing a loss of the correct source name in the TV displays. Only the first source name was displayed. Moved from Oz to 15APR91 and 15JUL91. 6506. February 8, 1991 AVSPC Phil The Australians needed an option in which AVSPC could apply an FG table while averaging. It now does this, the aim being that bad spectral channels can be flagged before a new channel 0 is created. Also updated the .HLP file. Added Neil's recent minor changes. Moved from 15JUL91 today. 6507. February 10, 1991 CVEL Phil Found and fixed the indexing error that was the cause of a long- standing bug: CVEL never worked properly for dual polzn data. That didn't really matter because none had ever been thrown at it in earnest until now. Added the change made recently in TST to correct MkIII data with the correct reference station. Also updated CVEL.HLP Moved from 15JUL91 TODAY. 6508. February 10, 1991 FRING Phil In the delay-rate determination subroutine if data containing only one baseline was passed and its SNR was below the threshold the matrix order for the LSQ routines was set to 0, this caused a divide by 0. Set a trap to catch this condition and just flag the entry in the SN table. Moved from 15JUL91 today. 6509. February 13, 1991 IBLED Phil LL only reared its ugly head again. When flagging single-source LL only data IBLED merrily passed all data through without flagging it because of the Stokes indexing error we fixed for most other tasks. Moved from 15JUL91 today. 6510. February 14, 1991 $INCIBM/DAPC.INC Glen Made the IBM version of the Pseudo array processor code consistent with the QPSAP version, except that the array size is 1.25 Mega- words=5 Mega bytes. Moved from 15JUL91 today. 6511. February 15, 1991 SYSAIPS:AREAS.DAT SYSUNIX:AREAS.CSH,SH Brian Updated these to have the IBM 3090 areas, APL3090, INC3090, Q3090, SYS3090. Moved from 15JUL91 this date. 6512. February 15, 1991 APL3090 INC3090 Q3090 SYS3090 Brian These areas were populated with the 3090 port done in Kingston and at Cornell. APL3090 doesn't yet contain the Double to IEEE and back conversion routines since IBM hasn't given them to us (an oversight probably). The INC3090 area just contains a 1MW DAPC.INC. The Q3090 routines are essentially the PSAP routines with vectorization directives as well as a library FFT. The SYS3090 area holds the appropriate LIBR.DAT etc. for the 3090. Moved from 15JUL91 this date. 6513. February 18, 1991 INSUNIX:INSTEP3 SYS3090:LIBR.DAT Brian Although some programs formerly in AIPPGM have migrated to AIPNOT, the changes hadn't migrated to the INSTEP3 scripts. A similar affliction affected LIBR.DAT. Moved from 15JUL91 this date. 6514. February 18, 1991 APL3090:ZR32RL Brian This made reference to S2VIBM rather than S2IBM. These are both IEEE to IBM local format routines in assembler and fortran respectively. IBM only donated the fortran version. Moved from 15JUL91 this date. 6515. February 19, 1991 INC3090 Brian Bumped up the "AP" size to be 1.25MW to be consistent with other AP's for DDT purposes. Moved from 15JUL91 this date. 6516. February 25, 1991 BPASS Phil When I redid the indexing for multiple IF's I forgot to do one loop that dealt with autocorrelation data. Athol's MKIII data uncovered the problem and its now fixed. Moved from 15JUL91 today. 6517. February 26, 1991 ACFIT Phil When confronted with dual polarization data, ACFIT fell flat on its face due to an indexing error in determining the number of channels. Another problem was uncovered which was due to the nature of MkIII data - it is possible to have a polzn or IF at a particular time which is all zeroes, ACFIT could not deal with that. The routine which read in the template file was only removing a zeroth order baseline from the template rather than the same order polynomial as was removed from the data, fixed that. Changed the philosophy of the program so that the user cannot select STOKES or IF's to be solved for, the program will now do all that exist in the file. May change this again later, but its a pain to sort out all the indexing at the moment. Added REFANT to the adverb list, so the template file can contain data from many antennas and the user just selects which is to be the reference antenna. Also changed the .HLP file. Moved from 15JUL91 today. 6518. February 27, 1991 CVEL Phil More indexing troubles, basically caused by the fact that CVEL reads and writes a multi-source file. Its very difficult keeping track of the correct JLOCx indices in the catalogue header when DGHEAD has changed the order in the output catblk. Hope I have it all sorted out this time. Moved from 15JUL91 today. 6519. February 27, 1991 FITLD Phil FITLD was not printing the important message that the file it was attempting to read was not a FITS file. It just died of unnatural causes. Added a call to MSGWRT. Moved from 15JUL91 today. 6520. February 27, 1991 SETJY Phil SETJY was obtaining the channel bandwidth from the SU table which does not cover the cases in which ch. bandwidth differs from IF to IF, or from FREQID to FREQID. Modified to use the FQ table if it exists, if not to use the catblk freq. increment. Moved from 15JUL91 today. 6521. February 28, 1991 CVEL Phil CVEL was suffering from the same afflication as SETJY (see entry 6520). Fixed this up. Also if no CL tables are present changed the code so it doesn't enter the routine designed to read them. Moved from 15JUL91 today. 6522. 2 March 1991 UVCOP Neil/Eric Commented out a seemingly useless and obstructive test for too many SELECTED channels (i.e. when the full band is not selected). I must be missing something or this is just wrong. Corrected another error : the task tested for FQ number even when there was none in the data set. Moved to 15APR91 and 15JUL91 from Oz. 6523. 2 March 1991 FUDGE Eric Raised buffer sizes. Move from Oz to 15APR91 and 15JUL91. 6524. 2 March 1991 PRTHI Eric Changed AU7 to display history-file line numbers when the printer or terminal is 80 or more characters across. Also brought the correction to RENAME done in CV to check for duplicate file names regardless of type. Moved from Oz to 15APR91 and 15JUL91. 6525. 2 March 1991 BLANK Eric Changed it to remember the zoom factor and center from frame to frame. Corrected argument to TVFIDL to use the correct number of TV levels (LUTOUT + 1). Moved from Oz to 15APR91 and 15JUL91. 6526. 2 March 1991 HISEQ Eric New task - converts image intensities to a histogram-equalized form. This is very good for display, but the intensities are then non-physical. Moved from Oz to 15APR91 and 15JUL91. 6527. 2 March 1991 Plot labels Eric Changed DLOC.INC, SETLOC, CLAB1, CLAB2, and IAXIS1 to display the epoch (B1950 or J2000) after the words Declination and Right ascension on plots. SETLOC puts the actual epoch from the header in to the DLOC.INC common where it is used by CLAB1 and CLAB2. IAXIS1 gets it straight from the TV image header. Changed PROFL as well. Moved from Oz to 15APR91 and 15JUL91. 6528. 2 March 1991 CHLORU Eric Simple subroutine to convert a string to all upper-case letters or to all lower-case letters. Moved from Oz to 15APR91 and 15JUL91. 6529. 2 March, 1991 TVMOVIE, SSS Eric Changed TVMOVI.FOR to correct the Y zoom position and, for FACTOR = 1, the Y scroll position and the X scroll (by 1 only). Changed $YSERV/SSS.SHR section CURSOR.C to restore the old cursor location computation and corrected the setting of the window parameters to include the translation from AIPS TV coordinates to those of SUNView (0,0 at top left). Changed section HEADER.H to increase the border by two pixels, decreasing the area to 1140 by 875. Moved to CV, without the Y size (they use a smaller one). 6530. March 6, 1991 CVEL Phil The fix I made on Feb 28 finally got compiled on kong, and I found out that I had not declared a loop counter. Bloody Convex compiler. Fixed. Moved from 15JUL91 today. 6531. March 7, 1991 FRING Bill C. Added an option invoked by APARM(5)=2 to fit for both multi- and single-band delays in the least squares solution. In this case the coarse search still fits a multiband delay. The SN table entries include the multiband delays in the IF dependent phase and the single band delays as the "delay". Also changed FRING.HLP. Fixed bug which alowed reference times for two polarizations to be different. This caused a phase error for the first polarization in the calibrated data of the fringe rate times the difference in reference times of the two polarizations. Enhanced diagonostic messages. Corrected the normalization of the weights that was allowing solutions well outside of the specified window. Improved the handling of baseline weights in FRNSRC. Bad data in one IF could cause all data for a given antenna to be discarded. Moved from 15JUL91 this date. 6532. March 7, 1991 CVEL Phil Dusted off a subroutine (whose call sequence had been commented out) which interpolated across flagged channels. Needed to do this because with MkIII data flagged channels are a reasonably common occurence. When MK3IN flags a channel zeroes are written into it, this causes a serious problem when shifting AC data after bandpass correction because the BP correction results in the value -1 being written into the flagged channel's value. When this occurs and the data are FFT'd and shifted the resultant spectrum is contaminated by ringing. Now interpolate across flagged channels before shifting and also shift the weights in a more correct manner. After shifting the channels that shift from one end of the spectrum to the other are now flagged. Moved from 15JUL91 today. 6533. March 8, 1991 $APGNOT/TBOUT.FOR Bill J. Prior to writing a table row to the output file, a test was made for FBLANKed values. The checks the value of the equivalenced REAL array for FBLANK. However, the equivalenced INTEGER array RESI may contain a valid record of -1 which has a REAL representation of -NaN and this causes problems. Added a trap for this situation. Moved from 15JUL91 today. 6534. March 11, 1991 APL3090: D2IBM D2IEEE Brian IBM finally got around to giving us the double precision IEEE to IBM and reverse routines (an oversight). Moved from 15JUL91 this date. 6535. March 13, 1991 $YSERV/XAS.SHR, AIPSSHAR Pat XAS.SHR didn't have a Makefile. Harry Payne kindly contributed one that I hacked on a bit and tested (mostly). Noticed some misleading comments and usage notice in AIPSSHAR while doing this, so fixed those too. Moved from 15JUL91 this date. 6536. 14 March, 1991 TVFLG, SPFLG Eric Changed SPFOAD, TVFOAD, DSPF.INC, DTVF.INC, SPFLG, TVFLG, and their help files to implement 3 new display modes. These are based on computing a running average of the data over a period of time arbitrarily called a "scan". The first new display is of the amplitude of the difference of the current sample with the vector scan average. The second new display is of the difference of the amplitude of the current sample and the scalar average amplitude in the scan. The third new display is of the difference of the phase of the current sample and the phase of the vector averaged scan. These --- especially the first --- provide powerful new ways to find discordant data without having to delete a full averaging interval as with methods using the rms and rms/mean. Corrected an error in the routine that finds the highest flag number in the flag control file. I had added a column of data (the scan time of the flag) and had not changed the pointer value. Fixed the code to get the pointer symbolically and hence to avoid this bug in future. Also fixed a minor addressing error in SSS --- routines CURSOR.C and SCRWRT.C. Moved from Oz to 15APR91 (so that they will be released) and 15JUL91. 6537. March 15, 1991 $SYSUNIX/FC Gareth Correct a couple of minor bugs in handling of options. NOOPT now removes all compile options including OPT1. Specifying a given option (e.g. OPT3) now removes all other options before adding the option correctly (it used to add the string "OPT3" instead of interpreting it). Moved from 15JUL91 this date. 6538. March 19, 1990 DECOR Bill C. Corrected call sequence to decorrelation routine. The visibility data were not being passed correctly. Moved from 15JUL91 this date. 6539. March 19, 1991 $APGNOT/DBCON.FOR Bill J. Check for compatibility of two files to be glued together made unwarranted assumption that the frequency axis increment was positive. This isn't the case for some WSRT data, in particular. Moved from 15JUL91 today. 6540. March 20, 1991 UVAVG Phil Added an option whereby cross and autocorrelation data can be merged seperately if their averaging times inside the data file are different. Also update the HLP file. Increased buffer size to 5 Mwords so that any VLA dataset can be averaged without sorting to BT. These changes are to bring the NEW version in line with the TST. Moved from 15JUL91 today. 6541. March 20, 1991 BPASS Phil Added another new mode to BPASS. For VLBI experiments one may often want to apply delay and rate calibration and then divide by channel 0. However for normal use we preclude this because users were often calibrating their data with the complex gains and then dividing by channel 0 which caused all sorts of problems. So now, if BPASSPRM(11) > 0 then the user can calibrate and divide by channel 0, however the input file tells the user not to do this if he/she does not understand what they are doing. Also changed BPASS.HLP, and then NEWPARMS.001 and POPSDAT.HLP because BPASSPRM grew. These changes are to bring the NEW version in line with the TST. Moved from 15JUL91 today. 6542. March 20, 1991 $QNOT/UVUNIF and UVTBUN Glen UVUNIF and UVTBUN now use the frequency defined by FREQG(CHUV1) for scaling the input uvdata before gridding. This was done so that UVUNIF and UVTBUN will use the save scaling factors as are used by UVGRID and UVGRTB. Moved from 15JUL91 this date. 6543. March 20, 1991 $INC/DAPC.INC Glen Another change to the comments. Now has values for the usual AP sizes. The work array RWORK has been moved to QCFFT. Moved from 15JUL91 this date. 6544. March 20, 1991 $APLNOT/UVZRWT.FOR Glen New subroutine that is used by UVGRID and UVGRTB.FOR. Moved from 15JUL91 this date. 6545. March 20, 1991 $QVEX/QXFOUR and QXFOR2 Glen Added two subroutines to QVEX, the old single precision QXFOUR and QXFOR2. These routines are faster on the convex than the four times more accurate double precision ones in $QPSAP. Moved from 15JUL91 this date. 6546. March 21, 1991 SNSMO Bill C. New task to filter SN tables. Much of the smoothing functionality of CLCAL is reproduced but this task included two useful model for smoothing FRING VLBI solutions. Coupled smoothings of delay (single and multiband), rate, amplitude and phase. The methods attempt to maintain coherence amoung the IFs of a given Stokes' parameter (but not between Stokes'). This task should be very useful for MkIII VLBI and VLBA data. Moved from 15JUL91 this date. 6548. March 22, 1991 TVFLG Phil The call sequence to ANTINI was missing a variable again! This is the second time I have fixed this (see entry 6488). This is one problem caused by testing things under a system that is not fully up to date. Moved from 15JUL91 today. 6549. March 22, 1991 CLCOR Phil LL only again, CLCOR will never have worked on LL only data. Added one extra line in the input parameter subroutine fixed it. Moved from 15JUL91 today. 6550. March 21, 1991 SNSMO Bill C. Fixed bug for some smoothing types in which a flagged gain was interpolated but the weight was left non positive effectively flagging it. Moved from 15JUL91 this date. 6551. March 26, 1991 IBLED Phil There was a problem when IQUV data were edited in IBLED. This occurred when the user restarted a session and IBLED read the FC table from the scratch file, it did not know how to deal with the 'IQUV' string in the polzn column. Fixed. Moved from 15JUL91 today. 6552. March 26, 1991 FRING Phil HAd an annoying little bug, when the user specified a SOLINT and FRING adjusted this downwards due to memory limitations it told the user but then went and wrote the solutions at the original interval. Changed 2 things: (1) Changed XASIZE from 100,000 -> 300,000 to lessen the possibility of this happening, should cope with most MkIII cases, but not all. (2) Set a trap in FRNGIN to check the solution interval (if non-zero) to see if it is viable, if not resets it at the very top level. Moved from 15JUL91 today. 6553. March 27, 1991 CSCOR Bill C. Fixed a sign error of the antenna longitude in calculating zenith angle and now converts longitude to sidereal units. Moved from 15JUL91 this date. 6554. March 27, 1990 SNCOR Bill C. Added new options: 'CLPA' Flag gains with amplitudes outside a given range. 'CLPP' Flag gains with phases outside a given range. 'ZPHS' Set gain phases to zero (useful for amplitude only calibration). Also fixed a bug in the handling of data with LCP only (yes another one.) Changed SNCOR.HLP. Moved from 15JUL91 this date. 6555. March 28, 1991 BPASS Phil/Neil Neil found a bug in the spectral interpolation routine which caused it not to work in all cases. Fixed. Moved from 15JUL91 this date. 6556. March 28, 1991 CVEL Phil Modified to determine the reference station from the baseline number as described in entry 6586. This removes the necessity of the user informing CVEL whether the data is MkIII or not, so changed the .HLP file to reflect this. Moved from 15JUL91 today. 6557. March 28, 1991 VLBIN Phil Added some extra adverbs to govern data selection. The user can now specify a bandwidth and frequency range to load into AIPS. Also the user can specify the reference frequency to be inserted in the catalogue header, this will override the first value found in the data - is useful when concatanating spectral line files. Modified .HLP to reflect these changes. To facilitate CVEL shifting of the data down the line, I now encode the reference station used in the correlator as the 1/1000th part of the baseline number (a la subarrays), this will work if there are no more than 20 antennas (not likely for MkII spectral line experiments). Moved from 15JUL91 today. 6558. April 1, 1991 TVFLG, SPFLG Eric Changed the scalar amplitude and phase difference displays to show the absolute value of the difference. The sign of the difference is of little import compared to the magnitude. Moved from Oz to 15APR91 and 15JUL91. 6559. April 1, 1991 XMOM Eric Corrected an error that had the image header for the 1st moment correct, but the image itself too low by the reference pixel value. This was an overhaul bug - the reference pixel value was hidden in the image scaling to avoid loss of precision which we now can no longer avoid. Moved from Oz to 15APR91 and 15JUL91. 6560. April 5, 1991 $APLSUB/SETLOC Pat Corrected a minor bug that only affected Aitoff projections when the reference pixel was non-zero. Moved from 15JUL91 this date. 6561. April 8, 1991 CLCOR.FOR, CLCOR.HLP Gareth Added a new OPTION, which I have called SSLO, to cope with the case of two VLA IFs which are normally locked together but observing at the same center frequency with different bandwidths. In this case, the frequencies are not locked together and the Signed Sum of the LOs used for phase tracking is wrong for one of the IFs. This mode should only used by experienced VLA spectral line users. For further confusion, see the help file. Moved from 15JUL91 this date. 6562. April 8, 1991 $APGNOT/UVPRM.FOR Glen Added a few new measurements. UVPRM returns the Observing Time, Min and Max Baseline length, Source Flux and Error, and reference antenna. Moved from 15JUL91 this date. 6563. April 8, 1991 $RUNSYS/MAPIT.001 Glen Improved use of ADVERB SOLINT, which is defined as the longest reasonable PHASE Only self-calibration interval. During the first 2 iterations, SOLINT is used in CALIB. In latter iterations SOLINT/2, SOLINT/3, SOLINT/4 ... are used. For Amp and Phase selfcal, the solution interval is 4 times SOLINT. Also Fixed help file MAPIT_MX, a list of MX adverbs which can also be used with MAPIT, but are not normally changed. Many Small improvements. Moved from 15JUL91 this date. 6564. April 8, 1991 $HLPFIL/MAPIT,MX,ZEROSP Glen Describe MAPIT in three sections, MAPIT_NA, MAPIT_UV and MAPIT_MX. Describe new MX zerospacing and uvweighting capabilities. Moved from 15JUL91 this date. 6565. April 8, 1991 Plotting Eric Corrected the Y position of vertical axis labels in CLAB1, CLAB2, TKLAB, PROFL, and GNPLT. Added the epoch to TKLAB coordinates. Corrected the positioning of the vertical label in LWPLA. Moved from Oz to 15JUL91 and 15APR91. 6566. April 8, 1991 POSSM Eric Corrected output to text file to label the columns (now including the IF) correctly. Moved from Oz to 15JUL91 and 15APR91. 6567. April 11, 1991 $QYPGNOT/MX.FOR Glen Added a collection of Multi-IF scaling tests. Also clean now stops at first negative CC if requested. Zerospacing flux now allows a source size parameter. Beam shape is fit by a gaussian, even if no components are restored. Moved from 15JUL91 this date. 6568. April 11, 1991 $QNOT/UVGRID,UVGRTB Glen Changes for multi-if data scaling. Moved from 15JUL91 this date. 6569. April 11, 1991 BSMAP Gareth An illegal Fortran statement (comparison between a COMPLEX and a REAL in an IF clause) caused the Sun compiler to fail when trying to optimize. Fixed. Moved to 15APR91 and 15JUL91 this date. 6570. April 12, 1991 SGINDEX Eric Corrected SG version number in AU2A to match that in SGLOCA - it's now 11 in NEW and 12 (for TVLINE) in TST. Moved from Oz to 15JUL91 and 15APR91. 6571. April 12, 1991 TVBLINK Eric AU6A did not reset the selected channel properly when the TV has no split screen (i.e., SUNs). Added an OFFF for the other channel to go with on ONNN for the desired one. Moved from Oz to 15APR91 and 15JUL91. 6572. April 12, 1991 $APLNOT/TABFRM Glen Modified TABFRM to allow "F" Format (i.e F5.2). Don Wells said we should be more generous on read than on writing FITS. Old IMLOD used to give IMLOD: TABFRM: BAD FITS TABLE FORMAT F5.2 But now "F" Format is interpreted as "E" Format. Moved from 15JUL91 this date. 6573. April 15, 1991 $APLSUN/ZABORS.C Gareth Added "#ifndef sparc" to make this module usable on both Sun 3s and Sun 4s/Sparcs. Tested on 4.1.1 on Sun Sparcs and on Sun 3s. Moved to 15APR91 and 15JUL91 this date. 6574. April 15, 1991 FILLM.FOR, FILLM.HLP Gareth 1. Fixed formatting bug in HI entry writing. This caused some input parameters not to appear in the HI file. 2. Minor fix in VLA band determination for the high end of L band. 3. Changed the definition of CPARM(7) (FQ tolerance) for continuum only. A value of <= 0 now forces a limit of 100 kHz. Added a message to tell user when all frequencies are being forced into the same FQ. This is wrong in several cases, e.g. when IF #2 changes but not IF #1. I would like to change the default, but I could not get a consensus from the spectral line users. 4. Changed the help file to reflect these changes. Moved to 15JUL91 and 15APR91 this date. 6575. April 15, 1991 $AIPSPUBL/COOK4.TEX Gareth Minor change to document change in FILLM. Moved to 15JUL91 and 15APR91 this date. 6576. April 15, 1991 POSSM Eric Corrected several axis labeling errors. The only significant one caused frequency and velocity axes to be offset by one channel. The F or v shown was for one channel higher than correct. Moved from Oz to 15JUL91 and 15APR91. 6577. April 17, 1991 POSSM Eric Corrected the code to plot (and print to OUTFILE) the correct velocities. Previously it was treating optical velocities as a linear axis. Further, it was putting "velocities" in the OUTFILE using the axis parameters (in part) even when they were in units of channels or MHz! The general belief is that this should not matter for small velocities. However, the differences amount to rather more than a channel when one has 1024 channels over 8 Mhz at 1420. Moved from Oz to 15APR91 and 15JUL91. 6578. April 18, 1991 $QYPGNOT/MX.FOR Glen Still trying to get the last few points in Bill J's data base rejected in MX for 4096**2, Two IF data base, when they fall outside the UV grid. Moved from 15JUL91 this date. 6579. April 22, 1991 $APGNOT/QUACK.FOR Bill J. Initialisation of antennas, channels and IFs for FG table was not being performed correctly using DATA statement on SUNs. Spurious values were introduced into the FG table --- usually very large numbers --- and LISTR et al. were trying to use this information and consequently being exceedingly slow. Moved from 15JUL91 today. 6580. April 22, 1991 $APGNOT/PRTUV.FOR Bill J. RA for output listing was being calculated incorrectly. Moved from 15JUL91 today. 6581. April 23, 1991 $APLPGM/PRTAN.FOR Bill J. Indexing in subroutine PRTLVA was incorrect with the result that pad N1 was being lost. This caused confusion for users. Moved from 15JUL91 today. 6582. April 18, 1991 $QYPGNOT/MX.FOR Glen Found reason for long lasting problems with MX data selection. UV grid for images is larger than for beams. (Clean beam is no bigger than 512 pixels). Because UV data can not be gridded near edge of grid the max UVrange is smaller for the beam than for images. Added a factor to reject a few more points near the edge of the grid. Moved from 15JUL91 this date. 6583. April 23, 1991 $SYSUNIX+$SYSSUN/LIBR.DAT Glen Adding $AIPNOT to LIBR.DAT files. Moved from 15JUL91 this date. 6584. April 23, 1991 $YPGNOT/IRING.FOR Bill J. Arguments in call to GINIT were incorrect. This caused IRING to fail if plot creation was required. Moved from 15JUL91 today. 6585. April 28, 1991 $SYSIBM/LINK Gareth There is a naming conflict in the C and Fortran libraries. With the newer versions of RS/6000 AIX (3003 and later - or whatever IBM numbers these releases now) the lbsd library and the lc libraries must be scanned before the lxlf (Fortran) library. Moved to 15APR91 and 15JUL91 today. 6586. 29 April, 1991 SYSUNIX:PP Brian In response to a complaint from Eric, I learned that PP has not been removing *.[fcs] files before preprocessing. This will cause preprocessing to fail if an existing file is not writable even if the directory is. Moved from 15JUL91 6587. May 5, 1991 $APLNOT/UVZRWT Glen Scaled zerospacing flux angular size for FWHM to sigma. Corrected logic error causing Zerospacing flux to be given a slightly too large weight. Moved from 15JUL91. 6588. May 6, 1991 TVFLG, SPFLG Eric FOUND a bug - TBTIME left the NX file open under LUN 30 whenever the user specified a single scan in his TIMERANG. Then it could not open the existing FG table as LUN 30 and so acted as if there were none. Corrected in both. Moved from Oz to 15APR91 and 15JUL91. 6589. May 8, 1991 $APLGEN/ZUVPAK Eric Corrected a stupidity --- the gain was set based on all data including flagged visibilities. Thus, much accuracy was lost to excessively high visibilities even when they were flagged. This might make sense if ZUVPAK had retained the vis values, but it set them to magic values in the very next loop. Moved from Oz to 15APR91 and 15JUL91. 6590. May 10, 1991 $APGNOT/FILLM Gareth There was a tape buffer resonance problem. When a correlator record ended precisely at the end of a physical tape record, all subsequent baselines for the current IF were offset by one word. This has been a problem since the program was first written. The problem occurs in 2-4% of all possible VLA permutations of spectral line modes and number of active antennas. It did not fail in continuum. Now fixed in the cases I could test - I hope I haven't broken other modes. Moved from 15JUL91 this date. 6591. May 12, 1991 $APLSUN/ZACTV9.C Gareth/Chris With revision 4.1.1 of Sun Unix, it was impossible to run AIPS tasks under the debugger. This was true on Sparcs, not on Sun 3s. It appears that the "wait" system call, used to wait for tasks in ZACTV9, is now obsolete. This has now been replaced by the "waitpid" system call, which works fine for both architectures. This Posix system call is not available on any version of Sun OS prior to 4.1. The generic version in $SYSUNIX will continue to work fine. If remote sites have these old revisions, they must remove $APLSUN/ZACTV9.C before running INSTEP2. Added to 15APR91 and 15JUL91 this date. 6592. May 14, 1991 TVFLG Phil For some inexplicable reason, 2 lines of code which used to exist in routine TVFMRK, and were essential to the flagging of single source files had been deleted. Put them back. Moved from 15JUL91 today. 6593. May 15, 1991 BPGET Phil There was an amazing screwup inside BPGET, someone (God knows who) had commented out an IF...END IF statement. This caused the routine to go and open and close the master scratch file for every visibility point, when normally this need only be done once. When fixed Junor and I got a speed up of a factor of 9! on a test data set run through HORUS. Moved from 15JUL91 immediately. 6594. May 15, 1991 DATBND Neil/Phil Neil fixed up the correction of IQUV and RL/LR data. Also the comments in the subroutine header were misleading. Moved from 15JUL91 today. 6595. May 16, 1991 $QNOT/MAKMAP.FOR Glen Too cautious, MAKMAP, error fixed in TST a month ago, but was afraid to move it to new. Scratch file was not being intialized after the first of 8 channels. Moved form 15JUL91 this date. 6596. May 16, 1991 $QPGNOT/CALIB.FOR Glen Added limits on the number of calibration sources printed and made the default CC table the highest. Improved History File text. Shortened HITIME.FOR printout. Moved form 15JUL91 this date. 6597. May 16, 1991 $YPGNOT/SNPLT.FOR Glen Allowed VLBI L or R only data with optyp= 'SUM', summary. Moved form 15JUL91 this date. 6598. May 16, 1991 $APGNOT/MOMFT.FOR Glen Moment Fit now checks for blanked pixels and excludes below a minimum input brightness. Moments are calcuated as the first, second and third power of the source brightness which gives a more reliable measure of compact source structure. Also changed MOMFT.HLP Moved form 15JUL91 this date. 6599. May 18, 1991 Workstation tape handling Gareth This is a fairly extensive change to tape handling mainly for workstations. First, since Unix workstations do not have facilities for tape device allocation, the usual mode of operation is "INTAPE n; GO xxxxx" with no "MOUNT" command. The mount command on workstations has previously made device assignments, but this is not necessary. It is easier to do and easier to control if the assignments for MT0n are required to be done in ASSNLOCAL.SH. So $APLIBM/ZMOUN2.C is removed completely. $APLUNIX/ZMOUN2.FOR is new. It always returns a successful mount/dismount condition. Second, remote tapes. I have packaged a public domain remote tape handling utility for use in AIPS. This allows a Unix machine to use a tape drive on another one if the basic tape I/O commands are identical. The remote tape capability is selected by defining the option REMOTE_TAPES in CCOPTS.SH. If it is selected, then I/O calls for drive functions are intercepted by the routines in ZRMTAP. The device name for the open is of the form: [:]/dev/ If the system name is absent, then all subsequent commands to the device are performed on the local system. If the system name is given, then a pipe is opened to that system, and all subsequent operations on that device are passed to the remote system. $APLBERK/ZRMTAP.C Public domain package with wrappers for AIPS. $APLBERK/ZTAPIO.C All of these routines will $APLBERK/ZTPCL2.C use the entries in the ZRMTAP $APLBERK/ZTPMI2.C modules instead of the native $APLBERK/ZTPOP2.C system calls if REMOTE_TAPES $APLBERK/ZTPWA2.C is defined. If not, they produce the same code as the versions in $APLUNIX. $APLSUN/ZTAP2.C These routines use the ZRMTAP $APLCVEX/ZTAP2.C entries if REMOTE_TAPES is $APLCVEX/ZTPOP2.C defined. Otherwise they $APLCVEX/ZTPWA2.C produce the same code as they did previously. This has been tested on Sun 3, Sun sparc, Solbourne, and Convex. It is likely that these are interoperable with DECstations, since the basic tape operations seem to be compatible. Alliants may also be compatible with the above; IBMs are not. However, it is likely that this will work between IBMs. Moved to 15APR91 and 15JUL91 this date. 6600. May 18, 1991 $APLNOT/UVZRWT.FOR Glen Modified zero spacing flux subroutine for MX. Now does not assume exact Int to Real Conversion and limits model for extended sources. Gaussian Model for extended sources is limited to only 1/100 of the peak (zerospacing) flux. Moved from 15JUL91 this date. 6601. May 19, 1991 $RUNSYS/DDTHLOAD.001 Glen Assorted fixes to speed execution and improve map. Changed CellSize for clean and fixed Test outdisk error. Moved from 15JUL91 this date. 6602. May 20, 1991 $QNOT/ALGSUB,ALGSTB Glen/Bill J. Found another large field multi-IF bug in MX. Added more logic to scale the Max U by the frequency. Fixed in 15JUL91 on April 1, 1990, but did not move it to NEW. Moved from 15JUL91 this date. 6603. May 19, 1991 SSS TV Server Eric Cleaned up the typing some more. Corrected the drawing of pixels when the image is zoomed to allow for partial pixels on the sides of the visible window. Changed the pixel addressing by one as well to get the hot spot of the cursor to be the actual pixel at its tip. It seemed to be one left and one above before. Moved from Oz to 15JUL91 and 15APR91. 6604. May 9, 1991 $APGNOT/ACFIT.FOR Athol/Glen Incorrect averaging of the last record in some scans. This record was sometimes corrupted. This effect was more noticeable for shorter solution intervals. Moved from 15JUL91. 6605. May 21, 1991 DATBND Phil After some discussion Neil convinced me that I was needlessly conjugating the imaginiary portion of the bandpass correction for cross-hands. Moved from 15JUL91 tpday. 6606. May