----------------------------------------------------------------------- ; 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 the master version of AIPS kept in Charlottesville. An entry 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 15APR92 *********************************************************************** Changes while 15APR92 was TST: *********************************************************************** *********************************************************************** 6739. August 1, 1991 QUACK Neil/Phil QUACK was writing channel and IF ranges as 0,0. However, UVFLG defaults BCHAN and BIF = 0 to 1, so that QUACKed entries in the FG table could never be unflagged. Changed QUACK to write channel and IF ranges as 1,0 Moved from Oz. 6740. August 1, 1991 SETJY Neil/Phil Add check for CALC and explicit FREQID setting A bug was introduced into GETFLX for calculation of 1934's flux density. ISRC was not correctly set. The HELP file needed minor chnages for 1934 and FQ selection with the CALC OPTYPE. Moved from Oz. 6741. August 1, 1991 DGHEAD Neil/Phil For Stokes selection IV, DGHEAD was writing the output Stokes axis increment as 4 rather than 3. Moved from Oz. 6742. August 1, 1991 RMEXT Phil Although advertised in the precursor comments as deleting all versions of a table if VER = 0, it didn't do so. Moved nowhere. 6743. August 1, 1991 MAKTAB Phil Minor modification so that if MAKTAB opened a table for writing and that table already exists it will not rewrite the column titles and units. Moved nowhere. 6744. August 1, 1991 R3DTAB Phil Minor modification so that if R3DTAB is writing to an already existing table it will append the new values rather than overwrite those that already exist. Moved nowhere. 6745. August 16, 1991 APLBERK/ZCREA2.C Pat/Gareth Remove reference to #include . None of the symbols in this file are referenced in the source and it was breaking Convex OS9.0+ versions of AIPS. Moved to 15OCT91 (and 15APR92, here) this date. 6746. August 23, 1991 New include files for binary tables Phil A host of new include files are needed to define the structure of the temporary binary tables for the VLBA distribution format. These tables are temporary in the sense that when FITTP writes this format it creates them, writes them and then deletes them. Also when FITLD reads them it creates the, transforms them into standard AIPS tables and then deletes them. The includes contain all the information needed to define a table in a general way. They are: DACV.INC - Antenna Characteristics table DAGV.INC - Array Geometry table DBAV.INC - Bandpass table DBCV.INC - Baseline Correction table DCAV.INC - Calibration table DFLV.INC - Flag table DFRV.INC - Frequency table DIMV.INC - Interferometer Model table DSOV.INC - Source table DUVV.INC - UV table DWXV.INC - Weather table Two other includes are also created. These are necessary for dealing with the FITS reading of the binary tables: DBHD.INC and for dealing with the global keywords attached to each binary table: DGLB.INC. Moved nowhere. 6747. August 29, 1991 BPASS.HLP Phil Craig pointed out to me that the explain file in BPASS.HLP was inconsistent with the help and inputs part of it. Tidied it up. Moved to 15OCT91 today. 6748. August 29, 1991 BPCOR Phil The option to correct for spectral indices across a bandpass correction was not working due to a coding error. Fixed and added another parameter to the APARMS for the SPEC option so the user could specify the channel about which the spectrum should pivot. Moved to 15OCT91 today. 6749. September 6, 1991 POSSM Phil The SOLINT option did not work correctly if SOLINT < 1. Fixed. Moved to 15OCT91 today. 6750. September 9, 1991 LISTR Dave A. Using DPARM(5)=0, Dparm(1)=7 with OPTYPE='GAIN' - if first scan was BLANK, would scale every scan to zero. Added some logic to subroutine GAIUV1 so code not only checks NUSCAN, but also if the scaling is zero. If it is zero, the scaling is recalculated until a non-blanked scan is encountered. Moved to 15OCT91 and 15APR92 today. 6751. September 9, 1991 $APLNOT/TABFQ.FOR Athol/Glen Array declaration Bug fixed. TABFQ declared a temp array assuming the FQ table had at most 4 columns (as the CH table does). Used existing parameter MAXFQC to specifiy the arrray. Moved to 15OCT91 this date. 6752. September 10, 1991 $RUNSYS/MAPIT.001 Glen Several complaints fixed. Now allows Non-square images, multi- field cleaning and use of FLUX for running MX. Moved nowhere yet. 6753. September 10, 1991 SETJY Phil The fix Neil and I put in on Aug 1 (6740) was too strong. Modified it so that if FREQID=-1 and OPCODE='CALC', the user now gets a very visible warning message instead of being thrown out of the task. Moved to 15OCT91 today. 6752. September 11. 1991 $RUNSYS/DDTHLOAD.001 Bill C. Added updated version of the procedures for running the Huge DDT test. They should also work for the smaller DDT tests. Moved nowhere. 6753. September 13, 1991 $APLIBM/ZLASC2 and ZLPCL2 Pat For some perverse reason, IBM AIX on RS/6000's appears to return from a system() call sometimes with a spurious error. The system call went fine, but "error 10, no child processes" would appear. Put in some "UGLY_HACK" stuff to ignore this. Yeah, this isn't the right thing to do but who has the time... Moved to 15OCT91 today. 6754. September 17, 1991 $QPGNOT/CALIB Glen Calib was failing when more than 1000 clean components were used in the source model for self-calibration. A buffer was too small and was being corrupted. Have now set XBFSZ=8192. It is hard to understand how it ever worked except that three scratch buffers were declared one-after-another. Added a test in $QNOT/VISDFT to make certain the scratch buffers are declared to be bigger than the minimum of 4096 words. Will move to 15OCT91 soon. 6755. September 17, 1991 $APGNOT/LWPLA Glen Modified the output so that it is more human readable. This is useful for plot editing tools like "pageview" and "ghostview". Also modified grey scale so that it is possible to have blank area inside a plot (not just very light grey). Also have lwpla skip printing if a grey scale line is all blank. (This speeds up greatly some plots.) Will move to 15OCT91 soon. 6756. September 20, 1991 $YPGNOT/KNTR.FOR Mark Added the capability to plot HPBW. Modified KNTR.HLP too. 6757. September 16, 1991 new FITS stuff. Phil Finally inserted all the new subroutines for dealing with the new VLBA FITS format into the system. Basically ecah new type of table has 3 subroutines associated with it, the first two characters of the subroutine name indicate for which type of table the routine is applicable. The 3 types of routine are: xxINI - routine to initialize the table for reading or writing xxIO - routine to read/write table keywords. xxTAB - routine to read/write table rows (does most of the work) Other routines that are part of this system are: GENKEY.FOR - to read/write a generic set of keywords TABERR.FOR - to write a standardized error message when an error is encountered. SETDEF/TABHDR - 2 routines which were modified to deal with the new format. Most of the work went into getting FITLD to read the new format. It has been extensively rewritten, but the new format is isolated as much as possible from the older formats. FITTP has also been modified in order to write this format, it is invoked by setting FORMAT=4. New FITTP.HLP is also present. Moved nowhere at all. 6758. September 20, 1991 FITTP Phil Putbck the old version of FITTP, Glen found that the catalogue header (number of complex axes) was changed to 3 if it was 1. Will fix. Moved nowhere. 6759. September 24, 1991 AHIST Eric New task: It does histogram equalization using a 2-dimensional window centered on each pixel. This allows local enhancement of structure in the image without regard for the intensity values in other parts of the image. My older task, HISEQ, does the equalization using the histogram of the entire cube. It is faster than the adaptive algorithm, but less effective. Moved from Oz to 15APR92. 6760. September 24, 1991 UVLSF Neil/Eric New task. Fits, by least squares, a straight line to the real and imaginary parts of selected groups of channels in a uv data set, and subtracts the fitted baseline from each spectrum. Also has the option to write a selected channel of the fit baseline as a separate uv data set. Moved from Oz to 15APR92. 6761. 26 July 1991 Multiple Eric Brought back several fixes from Mark, Neil and me: UVFLG Check error status returned by FLAGUP correctly so that IRET=-1 is not treated as an error (means last FG entry was read but it was already de-selected). TABED Increased all 1024 buffers to 4096. This so could merge BP tables with large numbers of spectral channels. MK3IN Changed to ignore error 10 from TAPIO and to mask length incompatibility messages. DATGET Make sure IERR is zero on input. BPASET Increased buffer size and declared it properly. Moved from Oz. 6762. 27 July 1991 TVFLG, SPFLG Eric Corrected the use of the previous cursor position when calling the enhancement routines; subscript (2,1) was used instead of (1,2). Changed the choices routine to reset zoom only when actually needed, rather than potentially needed. And removed confusing warning messages about scan lengths incompatible with averaging times. Messages will appear only when the incompatible parameter has had to be changed in order to use it. Move from Oz to CV 15JUL91. 6763. July 30, 1991 SPFLG, TVFLG Eric Changed the ``offzoom'' function to avoid YZOOMC calls unless required. Changed the code to leave graphics overlay channels 1 and 3 on and to clear out any lines and characters explicitly rather than calling YZERO all the time. This may be marginally slower on real TV devices but will be faster on workstations. Moved from Oz to 15JUL91. 6764. September 25, 1991 New task: AFILE Phil New task to manipulate the text A-files produced by the Haystack FRNGE program after it has run on the MkIII correlator output data. The A-file produced is a summary of the fringe-fitting performed on all scans correlated, it contains a large variety of information. Pertinent to this program are the time/date of correlation for a particular scan, the SNR of that scan, the "quality factor" of that scan and the length (in seconds) of the scan. Typically data is recorrelated so the user has to be able to tell MK3IN which copy of a scan he/she wishes to read into AIPS, this is done by generating a new A-file from the old, containing information only on those scans that are wanted. AFILE has the ability to sort the scan information by time and by baseline and also to edit it based on certain criteria specified by the user. The resultant A-file is then passed to MK3IN (fix on its way) and is used to select scans from the tape. Also a .HLP file. Moved nowhere. 6765. September 27, 1991 FITLD, *IO.FOR routines Phil Modified the handling of real/double precision global keywords. The IO routines and GLBKEY inside FITLD, had assumed that some were real and some d.p. John Benson had difficulty writing values with the D format (not a C construct) and so FITLD was bombing. Now check to see what type of data the keyword is and read it in the appropriate manner. Moved nowhere. 6766. September 30, 1991 DBCON Glen Seemed to be a problem with merging two multi-source datasets when running on an IBM 530. In second data set, the second sources was being renumbered as source number one. Moved nowhere. 6767. October 2, 1991 LWPLA Glen Found a compiler bug! (Apparently on several computers, but not baboon. The following text produces an error on the IBM 530: CHARACTER CHBUF*256 INTEGER NCHAR CHBUF = 'THIS IS A TEST' CHBUF = '(' // CHBUF(1:14) // ')' This assigns the value '((HIS IS A TEST)' to CHBUF. Created a temporary variable which produces the correcot result. Moved Nowhere. 6768. October 4, 1991 AHIST, UVLSF Eric Added a progress message to AHIST since it is slow. Added more history entries - the essential ones on CHANSEL - to both output files. Moved to new if possible. 6769. October 9, 1991 FITLD Phil There was a minor bug in routine REQCD - it had assumed that the OBJECT keyword would be close to the start of the file header. We had a case where this was not so and the task bombed because it skipped over essential keywords. Fixed so that this will not happen now. Moved nowhere. 6770. October 10, 1991 LISTR Phil Neil had missed (and therefore so did I) one instance of a call to CHNDAT when running LISTR in OPCODE='SCAN'. Fixed the call sequence. Moved nowhere. 6771. October 11, 1991 $APLNOT/CHNDAT.FOR Neil/Phil CHNDAT was writing, for the frequency increment in the FQ table, the signed header value. Thus, the true increment in the FQ table was lost whenever CHNDAT with the WRIT operation was invoked. I have added a new variable to the call sequence to pass the frequency increment table in and out of CHNDAT. This necessitates a lot of retro-fitting to AIPS as detailed below. I have also changed CHNDAT to return SIDEBAND=+1 when reading CH tables, as it always uses the signed header increment in this case. Moved nowhere. 6772. October 11, 1991 $APLNOT routines affected by CHNDAT change. Neil/Phil CALADJ Changed call to CHNDAT. CHNCOP Changed call to CHNDAT. Call sequence of CHNCOP changed to include work array for increment table passed from CHNDAT. DGHEAD Changed call to CHNDAT. FQFTAB A single precision array was being passed to CHNDAT instead of a double precision for the frequency offset table. I changed this, but can't find any piece of code that calls FQFTAB so it didn't matter. FRQTAB Changed call to CHNDAT. Removed some unused variables. GAININ Changed call to CHNDAT. Horrid use of FREQIF(1) and EQUIV (FREQIF,UBUFF) removed. Replaced DATP by ISBAND in call to CHNDAT for clarity. Removed unused variables GETFQ Changed call to CHNDAT. All the stuff reading the FQ table could be deleted and just let the call to CHNDAT deal with it now as CHNDAT and GETFQ return the same information now (except total bandwidth). Removed unused variable. SELSMG Changed call to CHNDAT. SN2CL Changed call to CHNDAT. Replaced SNREC by ISBAND in call to CHNDAT for clarity. Removed some unused variables SNSMO Changed call to CHNDAT. Moved nowhere. 6773. October 11, 1991 $QNOT routines affected by CHNDAT Neil/Phil MAKMAP Changed call to CHNDAT. UVMDIV Changed call to CHNCOP. Needed to add new work array to call sequence because of change to CHNCOP call sequence. Moved nowhere. 6774. October 11, 1991 $APGNOT programs affected by CHNDAT Neil/Phil BLFIT Changed call to CHNDAT. CLCOR Changed call to CHNDAT. Assume all IFs have same freq. increment in assigning BANDW, but BANDW is never used, as far as I can tell. CLSMO Changed call to CHNDAT. CSCOR Changed call to CHNDAT. FILLR Changed call to CHNDAT. In subroutine FLRCRE, assume both IFs have the same channel increment. This is what the old CHNDAT would have done, done, and I see no easy alternative given the structure. LISTR Changed call to CHNDAT. SDCAL Changed call to CHNDAT. Note the following bug which I have not fixed. FREQO(*) is an array passed into subrouitne SDCCOP as an input parameter. It is, as far as I can see, neither filled nor initialized. It is then added to the regular IF frequency offset. I presume it is meant to be the source specific offset. It is NOT in any common. I leave it to NRAO to do something with this. I also removed some equivalences used in SDCCOP SETJY Changed call to CHNDAT. CHNDAT now returns all (except total bandwidth) that GETFQ does. Remove calls to GETFQ, in subroutine SETJ in in favour of CHNDAT. Removed some unused variables in SETJ. Includes fix to warn users if FREQID not set by user for 'CALC' Includes fix in GETFLX with source indexing for 1934-638 SHOUV Changed call to CHNDAT. Replaced FREQIF with explicit use of FOFF(BIF) to make it clear in code that this offset is from the first IF only. SNCOR Changed call to CHNDAT. Removed unsued BANDW. SNSMO Changed call to CHNDAT. SPLIT Changed call to CHNDAT. In subroutine SPLCOP removed unecessary equivalences (ISBAND,VIS) and (FOFF,UBUFF). Included kludge for AT with multi freqid data bases. UVCOP Changed call to CHNDAT. Completely rewrote the section to copy FQ table information in subroutine COPYIN by using only CHNDAT. Deleted subroutine FQSCOP which is no longer needed. This was writing out the wrong part of the FQ table when IF selection was invoked, anyway. Deleted some unsued variables in COPYIN. Initialized PRTCNT in COPYUV. UVFIT Changed call to CHNDAT. Loop in UVFDAT to work out averaged freq. uses header channel increment. Replaced by FINC(LOOP) Perhaps FINC(BIF) would be safer? UVIMG Changed call to CHNDAT. There are assumptions throughout about equal frequency increments for the IF axis. Instead of using the header frequency increment, as a minimum change use the increment for IF number BIF from the selected FREQID. UVMOD Changed call to CHNDAT. Moved nowhere. 6775. October 11, 1991 $QYPGNOT programs affected by CHNDAT Neil/Phil MX Changed call to CHNDAT. Just before call to CHNDAT, there is a factor, GCOR, that is set with the header frequency increment. I don't know if this is meant to be absolute or not, so it is unclear whether it is should be using FINC or not. Also, the case of averaging two IFs complicztes this. I think the use of the correct frequency increment in the gridding procedures needs to be examined carefully (Glen ?). The frequency OFFSET is multiplied by the SIDEBAND indicator in subroutine MXSEL in caluclating IFFREQ. Removed the sideband indicator from this line. Changed call sequence to CHNCOP. Moved nowhere. 6776. October 11, 1991 $YPGNOT programs affected by CHNDAT Neil/Phil POSSM Changed call to CHNDAT. After first call to CHNDAT, use FINC to set a frequency scale factor. BUt can select more than one IF, so this only valid if all IFs have the same channel increment. In the last section of LABVEL, BANDW is used as the channel increment. This is fished out of the source table. For AT data, BANDW would be zero, as we don't fill it. SHould BANDW really be FINC(IF) ? This is unclear to me. SNPLT Changed call to CHNDAT. VBPLT Changed call to CHNDAT. Pass FRQSEL into subroutine ANTIN to get correct FQ/CH table info and correctly work out frequency of selected channel/IF/FREQID Looks for frequency in ANtenna table first. If zero then uses header reference freq. Moved nowhere. 6777. October 11, 1991 $QPGNOT programs affected by CHNDAT Neil/Phil ASCAL Changed call to UVMDIV. BLCAL Changed call to CHNCOP. Changed call to UVMDIV. BPASS Changed call to CHNDAT. Frequencies were computed by multiplying FQ offset by side band indicator and adding to the reference freq. Removed the side band indicvator from this. Call to BPCLER with 14th argment, GWORK(2*MAXANT), a double precision variable, integer expected in COUNT(MAXANT), but nothing is done with it, so who cares. Changed call to CHNCOP. Changed call to UVMDIV. Includes MRC's pointer fix for uncompressed data in routine EXTDIV which divides by channel 0. A data indexing variable CMPNT4 was not initialized for data in uncompressed format. CALIB Changed call sequence to CHNDAT. Put ISBAND and FINC in common Changed call to CHNCOP, used FINC, FOFF and ISBAND in call instead of faceless work arrays Changed call to UVMDIV. CVEL Removed unecessary include files from main routine. CVELIN: Remove calls to FNDEXT and GETFQ. Close FQ table if opened with FQINI. Remove call to CHNDAT. Removed now unused variables associated with unecssary calls. Removed IUDISK and IUCNO in favour of DISKIN and CNOIN CVSHFT: Move call to CHNDAT outside of loop and change call sequence. The desired FREQid has been selected by UVGET so it will always be FRQSEL. Test for ILOCFQ>0 not needed as CHNDAT will return whatever it can find, if no FQ table. DOSHFT used before it was computed in SHIFT. Move computation of DOSHFT into CVSHFT out of SHIFT. Removed out of common into local variable of CVSHFT. Removed FRQIF1. It was unused and computed with an unset variable (OBSFRQ). Fix calls to CHNDAT. COrrect output frequency offsets in WRIT call to CHNDAT. They are pretty much meaningless anyway after CVEL has move the spectra up and down. SHIFT: Removed unused variable PRDELI. Initialize HELIO and RADIO for single source files. Get frequency increment from FINC (from CHNDAT) regardless of whether single or multi source GETFRQ: REFFRQ, the frequency of the reference pixel of each IF was was being filled only for IF number one in the single source case. Same for the velocity of the reference pixel In multi-source part of code, loop over NUMIF rather than MAXIF Removed computed but unused variable TBW from CVEL.INC. Removed now unused variable NCHAN. The formula for VELINC comes from EWG's memo which has left out the minus sign for the radio convention. Added it. CVELHI: Added FQ table to list not to copy with ALLTAB VELPIX: the formula for VELINC comes from EWG's memo which has left out the minus sign for the radio convention; removed the unecessary ISIGN. FRING Changed call to CHNDAT. Put ISBAND and FINC in common. Changed call to CHNCOP; used FOFF, ISBAND and FINC arrays instead of faceless work arrays. Changed call to UVMDIV. GRIDR Changed call to CHNDAT. Remove BANWID in favour of FINC(BIF) PCAL Changed call to CHNCOP. Used (small) arrays with meaningful names for clarity. UVSUB Changed call to UVMDIV. VSCAL Changed call to UVMDIV. Moved nowhere. 6778. October 11, 1991 $APL/CONTRIBU programs affected by CHNDAT Neil RDCAL Changed call to UVMDIV. Moved nowhere. 6779. October 15, 1991 $QNOT/UVTBUN.FOR Glen/Ralph Gaume UVTBUN was not correctly calculating UNIFORM weighting for images as big as 1024x1024 or larger, because part of the AP grid used was not being initialized correctly. Simple change to Zero grid before calculations. (Note there was no problem for continuum obs or for 512x512 or smaller images) Moved Nowhere. 6780. October 16, 1991 Screen servers Eric Changed $YSS version of YWINDO to issue an Interogate parameters call to the server when there is a window size error and then reset the ID file's parameters if the result clears up the problem. Changed $YSS/YINIT to make this call automatically as well. Changed $YSERV versions of XAS.SHR, XVSS.SHR, and SSS.SHR to support the interogate function, which was not previously implemented. Also changed XAS.SHR to remove an error put in whereby the window resize function tested its location but did not then fix it --- making thereby an infinite loop at least on IBMs. Also changed XAS and XVSS to reposition the window when the resize button is pressed or YWINDO is called with write, but not on repaint interrupts. This allows users to move the TV off the screen if they want. Moved nowhere. 6781. October 16, 1991 $APLNOT/SN2CL Glen/Paddy Paddy at Jordrel experienced a problem with SN2CL when SN table records were duplicated. He sent the fix and I incorperated it into SN2CL. Linked CLCAL and ran VLAC successfully after applying the fix. Moved Nowhere. 6782. October 17, 1991 IMLIN, UVLIN Tim/Juan Image and Visibility based continuum subtraction programs. UVLIN is roughly the same as UVBAS but fits in real and imaginary parts of the visibility and hence has much better performance. It can also find and flag narrow-band interference by looking at the residual of the fit. IMLIN is the image-plane analog of UVLIN and works as well. UVLIN and IMLIN are the tasks described in the preprint by Cornwell, Uson and Haddad, Oct. 1991 (submitted to Astron. Astrophys.) Moved Nowhere. 6783. October 21, 1991 $APGNOT/LWPLA Glen Further messing with LWPLA so that I can use "pageview" to edit output postscript files. Font is gone up 1 size! Bigger font is always needed for publication. Moved Nowhere. 6784. October 21, 1991 $RUNSYS/MAPIT.001 Glen Several features including "OLAF-LIKE" interactive mode. Setting DOINTER=1 before executing MAPIT causes the procedure to ask the user to set boxes before further cleaning. Moved Nowhere. 6785. October 22, 1991 Sun Exabytes Eric/Mark Added Mark's (AT) code to the system. Changed ASSNLOCAL.SH and ASSNLOCAL.CSH in $SYSUNIX to set TAPEn logical to $MT0(n-1) for now. Mark's solution is better but needs many fixes. Changed Baboon's $SYSLOCAL versions of these --- other systems will need this (by hand???!!). Mark's comments: --------------------------------------------------------------------- 0000. 7 June, 1990 Exabyte Z routines Eric/Mark Calabretta Changed: APLSUN/ZMOUN2.C Created new routine to assign the logical MT0n to a device "/dev/nrstx" APLSUN/ZTAP2.C Added magic beginning of tape value and a test for it to support the Exabyte. APLSUN/ZTPCL2.C Added mask for errno=1 ("not owner") which the exabyte driver returns erroneously. Moved from Oz 17-June. 0082. 90/Sep/04 $APLSUN/ZMOUN2.C Mark Rewrote $APLSUN/ZMOUN2.C from the bottom up. It now uses lock files to prevent a tape being mounted simultaneously by different AIPS users. Locking is implemented via lockf() on files with names of the form $DA00/TAPE1.lock. This supplements the $DA00/TP* lock files which give a tape task exclusive access, but only while it is executing. The lock files are created on MOUNT, and deleted on DISMOUNT. $APLSUN/ZMOUN2.C now also uses environment variables TAPE1, TAPE2, etc. to define tape devices. TAPEx corresponds to INTAPE=x, and OUTTAPE=x. The TAPEx environment variables are defined in HOSTS.CSH and so can be tailored to each SUN workstation. They may be left undefined for workstations which do not have an attached tape drive. On mounting, the value of MT00 is taken from TAPE1, MT01 from TAPE2, etc. Initially, and on dismounting, MT0x is defined as "UNMOUNTED". It may be worth making this scheme more AIPS-generic by modifying ZPHFIL to return an AIPS-style file name, and also by using ZLOCK to manage the lock. Differing tape densities would need to be supported by adding a density suffix, e.g. TAPE1_6250. 0109. 90/Nov/02 $APLSUN/ZMOUN2.C Mark Changed to use flock rather than lockf as above. Should be changed to use ZLOCK when this mess gets sorted out. 0170. 91/Jan/18 $LOCAPLSUN/ZMOUN2.C Mark Now takes the Exabyte offline (ejects it) when it is dismounted. 0190. 91/Feb/11 $LOCAPLSUN/ZMOUN2.C Mark Modified to apply an exclusive lock rather than a shared one. 0125. 90/Dec/05 Exabyte mtio problem under SunOS 4.1 Mark The mtio magtape interface in SunOS 4.1 appears to contain a bug. The command 'mt -f /dev/nrst0 bsf N' (where N is a number) skips backward over N+2 tape marks instead of N. Remedied this by modifying $LOCAPLSUN/ZTAP2.C to use the new "MTNBSF" opcode which backspaces to beginning of file, and does work properly. Also modified $LOCAPLSUN/ZBKLD2.FOR to use "nbsf" instead of "bsf" in the mt command. Recompiled all tape tasks for the SPARCs. 0150. 91/Jan/09 $LOCAPLSUN/ZTAPE.FOR Mark I had modified ZTAPE on Nov/08 to write just one EOF, rather than write 4 and backspace 3. This was done because of the bug in the MBSF operation in mtio for the Exabyte which at that stage was not properly understood (see entry 125 above). However, FNDEOT relies on finding at least 2 EOFs at the logical end of tape, and in particular, FITTP was failing when it tried to append files. Rather than revert to the original ZTAPE, for speed reasons I modified it to write 2 EOFs and backspace 1. This saves about 25s in EOF processing per file. 0236. 91/Apr/16 SunOS 4.1.1 Exabyte rubbish Mark More Exabyte lunacy, this time occasioned by the fact that if the Exabyte is already at logical EOM, giving it an MTEOM causes it to go screaming off to the physical end of medium! Modified ZTAPE to recognize an EOM command, and ZBKTP1 to use it. Previously ZBKTP1 was attempting to 'mt -f /dev/nrst0 eom' via ZSHCMD while the tape was already opened by BAKLD. This used to work under SunOS 4.1 (I think) but under SunOS 4.1.1 it (sensibly) produces a 'device busy' error. It also doesn't allow for the possibility that the Exabyte is already at logical EOM. Modified ZBKLD2 to specify the tape drive explicitly in the MTEX command. Added an EOM operation to ZTAP2.C for ZTAPE as described above. The EOM first backspaces one file to ensure the tape is not already at EOM (there's no way of finding out from the mt status information!) Completely revamped ZTAP2 so that all ioctl status returns are checked. This is particularly important for the EOM command. In the past ZTAP2 could fail unnoticed leaving the tape in the wrong place, usually BOM. This could be disastrous if the intention was to append to the end of the tape. On the plus side, MTBSF now works properly, so removed the MTNBSF kludge from ZTAP2. Conclusion: the SUN Exabyte driver is an appalling piece of rubbish. (A rather bad bug in ZBKLD2 was not fixed at this time. The shell command line to "delete old files if any" i.e. TEXT = 'rm ' // ZDA0N(1:LZDA0N) // '/???' // CCNO // '???' // * CUSER // '??' deletes the message file and all save/get files if CCNO = CUSER!) 0243. 91/Apr/21 $APLSUN/ZTAP2.C Mark Added an escape clause for MTBSF with the Exabyte positioned in file 0. The ioctl returns errno=5 (IO error)! --------------------------------------------------------------------- 6786. October 23 UVHGM Tim Added X AXTYPE to allow plotting of vis*sqrt(wt) which is useful for evalating noise levels. Moved Nowhere 6789. October 30 UVLIN Tim Same definition of shift as UVFIX. Moved nowhere. 6790. November 6 $YPGNOT/VBPLT.FOR Athol VBPLT did not plot correct closure phase models. Various corrections in subroutine MODSET to achieve this. Moved nowhere. 6791. November 7 $APGNOT/SETJY.FOR Dave SETJY would not bomb for an invalid OPTYPE. Added an IF statement to test OPTYPE, and return if not valid. Note if more OPTYPE options are added to SETJY, this IF statement will need to be changed. Moved nowhere. 6792. November 7 $YPGNOT/GAL.FOR Glen Fixed printout, added warning when wrong type of image supplied and anotated the plot file. Made a few improvements to GAL.HLP Moved nowhere. 6793. November 12, 1991 DSEL.INC, BPGET Phil The bandpass applications routines could not deal with VLA data with a large number of channels (> 256). This was a consequence of the previous change to BPGET in that I had assumed that all BP data could be read in one fell swoop with UVDISK. This turned out not to be the case. Fixed BPGET so it now does read all the data in one go and stores the BP array in memory - the alternative is many opens of the scratch file, that causes me indigestion. Increased the size of the BP memory buffer to deal with the AT worst case. Moved nowhere. 6794. November 13, 1991 GAININ Bill C. Change several array declarations of (1) of equivalenced values to the actual dimensions. Moved nowhere. 6795. November 13, 1991 DATGET Bill C. Corrected the declarations of the arrays used for reading the index table to use the number of columns declared in PUVD.INC rather than the hardcoded and now incorrect value of 6. Moved nowhere. 6796. November 14, 1991 VLAL files Dave A. Put VLALEXEC.001 and VLALLOAD.001 into the RUN area in TEST; VLAL.HLP and VLALSAVE.HLP into the HELP area. These are the files necessary to run the Spectral Line DDT. Moved nowhere. 6797. November 15, 1991 $YSERV/XAS.SHR Glen Modified XAS to track which image and graphics planes have been written to. Clears to already clear planes are much quicker. Speeds GRCL if GRCHAN=0 and TVPL for clear TVs. Should also help IBLED. (Changes made expecting experineced TVers may make further modifications/improvements/corrections...) Moved Nowhere. 6798. November 15, 1991 CSLGET Bill C. Fixed a couple of equivalenced variables dimensioned (1). Moved nowhere. 6799. November 20, 1991 $APLNOT/GAININ Glen Fixed GAININ decoding of Column type and locations if the Column does not exist. Moved nowhere. 6800. November 20, 1991 $RUNSYS/MAPIT.001 Glen Incorperated suggestions of Ken Kellerman and Athol Kemball into MAPIT. Major Change was use of IN2N, which is now the input model for the First Self-Cal. If IN2N='POINT' then a point model is used, else IN2N is the name of the clean image to use. Improved MAPIT.HLP. Moved Nowhere. 6801. November 20, 1991 $APLNOT/DATCAL Phil When setting the logical to decide if it is necessary to perform delay/rate corrections. DATCAL just checks to see whether any corrections need to be applied to IF #1. In the case where IF#1 is flagged it will then fail to correct any of the IF's of a visibility spectrum for the residual delay and rate. Fixed so that this does not happen. Moved nowhere. 6802. November 20, 1991 $APLNOT/GETANT Phil Fixed a longstanding annoyance. The Block 2 correlator writes an AN table for all antennas in an experiment, however if there are no fringes to a particular antennas its coordinates are set to 0. This caused GETANT to blow up, even though the antenna in question was never present in the data being examined. Fixed so that this will not happen. Moved nowhere. 6803. November 21, 1991 $APLNOT/TABFQ Phil When writing FQ tables the array needed to hold each row was not declared to be large enough, especially for 14 IF's. I'm surprised this didn't catch us before, but it seems to work as is on a Convex, it fails on a Sun. Maybe nobody has pushed 14 IF data through before. Moved nowhere. 6804. November 22, 1991 MK3IN.FOR, MK3IN.HLP Phil Added the capability to select MkIII scans based on the Afile text information generated at the correlator. IN2FILE specifies the A-file to be used for the data selection (see entry 6764), if this is blank MK3IN will accept all scans, if this is non-blank MK3IN will only accept those scans whose root is valid. Also extened the maximum number of baselines from 45 to 120. Updated help file. Moved nowhere. 6805. November 22, 1991 GREYS Eric Put in a missing "GO TO 999" which caused an error message with IERR = 0 to appear and worry some users. Moved nowhere. 6806. November 22, 1991 $APGNOT/LWPLA Glen Found there was indeed a bug for grey scale plots for images wider than 256 pixels. A new line was inadvertently added every 256 pixels, which confused the printers. After great labor found that numbers could be aligned if the "SPACE" character was made wider in Helvetica-Bold by 14 units. Moved nowhere. 6807. November 25, 1991 QPGNOT Phil BPASS had a problem when running in autocorrelation mode on 1-baseline data. The 2nd antennas's data was never normalized because the normalization loop was over number of baselines, not number of antennas. This was the only case in which this would happen because in all other cases # baselines > # antennas. Moved nowhere. 6808. November 25, 1991 $SYSUNIX/ZLASCL Glen Added logic to use SUN's Pageview or GNU Ghostscript for Post Script files, if requested. Uses two addtional files ZLASCL.PAGEVIEW and ZLASCL.GHOSTSCRIPT. Moved nohwere. 6809. November 26, 1991 $APLNOT/BPGET Phil I don't believe this one. A user complained to me that DOBAND 2 did not work on her data, she showed me plots generated on the Convex that proved this. I fired POSSM up on a Sparc to verify this and it fell flat on its face with a bus error. The call sequence to SCLOAD inside BPGET was wrong. What worries me is that the Convex did not trap this, it just did the wrong thing. Fixed the call sequence and all is OK. Moved nowhere. 6810. November 27, 1991 $APGNOT/FILLM Gareth I find these hard to believe too! Fixed 3 bugs. 1. MAJOR PROBLEM. The reference frequency was changed with each new scan. Thus the (u,v,w) are not, in general, referred to the reference frequency in the catalog header! Frazer Owen is credited with finding this effect, which is now fixed. 2. MAJOR PROBLEM. When changing UV files, the CL table entry at the beginning of the next scan was not written! This can lead to some significant problems with applying calibration. 3. The total bandwidth were incorrectly determined for the narrowest back-end filters/recirculators. The first two problems are, of course, correctable using UVFIX and INDXR, but how many people did that? Moved nowhere. 6811. December 4, 1991 $APGNOT/LWPLA Glen Alan B. found a bug in LWPLA (in OLD, NEW and TST, ie I did not introduce it) which caused it to draw spurious lines when plotting a long slice. Found that yet another buffer was overflowing when trying to write 4096 characters at once. Moved nowhere. 6812. December 5, 1991 $YPGM/ISPEC.FOR Dave A. Program was not clearing the READ flag, although the routine was ending successfully. Removed two lines which were setting FRW in subroutine ISPCIN. Moved nowhere. 6813. December 5, 1991 $APGNOT/FILLM Gareth During July and August 1990, the VLA IAT clock was misbehaving. Unfortunately, this was the first test of the on-line software to handle clock errors; the software also failed. This led to total delays of the order of 1.0D50 nanoseconds being recorded on the VLA archive tape. Processing of these large numbers caused FILLM to die on Convexes and to produce nonsense results on Suns. Out-of-range delays now cause all data to be flagged bad. Moved nowhere. 6814. December 11, 1991 POSSM Phil There was an error when labelling plots from multi-source files in velocity units. The subroutine responsible assumed that the reference pixel was always 1.0 - what amazes me is that nobody had complained about this, why do the AIPS group have to run tasks on their own data before uncovering these irritating little bugs. Moved nowhere. 6815. December 16, 1991 BPASS Phil Made several minor changes to BPASS based on comments from Neil. Also fixed one outstanding bug, a remnant from the day this was written, but which fortunately people never really needed or ran into, in most cases. Minor fixes: (1) added a parameter (MXENTR) to specify the maximum number of time entries, true max = MXENTR * NANT (2) If there is no NX table for the multi-source case, BPASS will tell you and die. (3) Modified the logic slightly in the case of SOLINT -1, there was a bug which caused the NX table to be opned and never closed, even though it wasn't needed. (4) Modified the handling of the reference antenna to deal with the case in which there are different reference antennas in LL and RR (or XX and YY). Major fix: When the reference antenna changes during the observation BPASS was supposed to reference the output table so that all entries would have the same reference antenna. This option never even pretended to work, it now does. Luckily this was only necessary for the DOBAND = 3 option (interpolation of bandpasses) which people never really used. Some now want to so this is a first step towards making that option work properly. Moved nowhere. 6816. December 18, 1991 CVEL, CVEL.HLP Phil Received some changes from Neil, also made some of my own. (1) Added a new adverb, GAINUSE, so the user can tell the task which CL table to look in to determine the time/antenna dependent frequency offsets. Also updated the .HLP file. (2) CVEL now gives a warning to people who run it on single- source files, for the reason that they may not have the time dependent information in the CL table. (3) Tidied up the way CVEL deals with APARMS and how it writes them to the ALTxx words in the output catalogue header. (4) 2 calls to CHNDAT used the wrong LUN (5) Only close CL table if it was open (6) Rationalized the determination of the vel. at first pixel and shift calculations (7) Reinitialize ALLFLG after it has been set true by SPINTP Moved nowhere. 6817. December 18, 1991 LWPLA Glen Attempted to re-center the plot on the middle of the page. Moved nowhere. 6818. December 18, 1991 ACFIT Athol/Phil Several modifications to ACFIT. Corrected mishandling of the first point in each solution interval. Equalise the solution intervals and reject intervals with too few points. Non self-cal SN time sampling is justified by the large solution intervals typically used. Include an option to allow a residual polynomial as part of the fit to the emplate spectrum. This absorbs terms due to the source spectrum and the template spectrum and it is not neccessary to provide off-source channel ranges for these spectra (BPARM and CPARM). This method reduces systematic gain errors caused by baselines that change throughout an observing run. The new method is selected with APARM(9) (baseline-independent fit) and is useful for narrow- bandwidth observations. Also changed ACFIT.HLP. Moved nowhere. 6819. December 20, 1991 APLNOT:GRIDTC Phil When GRIDTC is building up the SPFLGR mapfile it does it in groups of baselines, the size of the groups depending on the number of channels/IF's/Stokes etc. If, for any reason, the first group, or any group, contained no visibilities, then SPFLG would fail with an obscure error message. Fixed so that this will not happen. Moved nowhere. 6820. December 20, 1991 AFILE Phil 2 minor problems that caused the last 1 or 2 lines of the output file to be output in error. Changed a .EQ. to .GT. when determining if past the total number of input lines. If last line had no pair with which was compared the check on experiment number was not done. Moved nowhere. 6821. December 23, 1991 FRQTAB Phil Fixed a small residual error from the NEBK CH/FQ changes. If there was no IF axis, FRQTAB did not call CHNDAT and therefore did not fill in FINC, which therefore caused the array FREQG (used in many places), to be all zero, except for the first element. Moved nowhere. 6822. December 24, 1991 BPINI Phil Discovered a case (MkIII data), where the existing buffer size declaeration was not large enough. Doubled it. Moved nowhere. 6823. December 30, 1991 Remote tapes Eric In order to provide a more robust method to handle remote tapes, I have created a ``deamon'' task called TPMON and revised Z routines etc.~to support it. In the non-Z areas, I have changed $INC/DDCH.INC Added parameters to keep track of mounted tapes. $INC/DAPL.INC Added adverbs REMHOST and REMTAPE for remote mounts. $HLPFIL/POPSDAT.HLP Added adverbs REMHOST and REMTAPE for remote mounts. $HLPFIL/REMHOST.HLP New adverb to specify the host name for the real tape. $HLPFIL/REMTAPE.HLP New adverb help for remote host tape drive number. $AIPSUB/AU4 Changed to call ZMOUNT with new parms rather than ZTAPE. Changed to avoid an excess back-file for FITS tapes. $AIPSUB/FWRITE Changed to assume that the first logical record is already available in the input buffer. $AIPSUB/UWRITE Changed to back-file for non-FITS data and reset the tape control parameters before reading. $APLSUB/TPHEAD Changed to allow more than 1 tape record to be read. $AIPNOT/GRITP Changed to use ZMOUNT rather than ZTAPE and to allow remote tapes. $AIPNOT/GR2TEX Changed to use ZMOUNT rather than ZTAPE and to allow remote tapes. $AIPNOT/TPMON New deamon to manage real tapes locally reading from a network commands to perform. TPMON1 is for FITS-disk files while TPMON2 - TPMON(m+1) are for tape drives 1 - m. Moved nowhere. 6824. December 30, 1991 Remote tape Z routines Eric Changed in $APLGEN the following: ZDCHIN Initialize the mount names to blanks. ZMOUNT New call sequence to include remote name and tape number. Then translates TAPEn logical and assigns AMT0n to it. The it calls ZMOUNR or ZMOUN2 as needed. On dismount it first calls ZMOUNR or ZMOUN2 and then reassigns AMT0n to UNMOUNTED. This routine will now always be used. ZMOUNR Opens the socket, sends the mount/dismount command, and closes the socket (for remote tapes only). ZCRLOG Dummy version of this create/reassign logical subroutine. ZTAPE Removed mount and dismount operations, added test and call to ZTAPR, cleaned up typing. ZTAPR New routine to pass ZTAPE arguments across the network. ZTPOPN Changed to call ZTPOPR for remote tapes. It detects a remote FITS-disk file via a double colon in the name. A mount will have had to be done since the logical name AMT0n is now used. ZTPOPR Opens socket and passes ZTPOPN arguments across the net. ZTPCLS Changed to test TPNAME from common and call either ZTPCL2 or ZTPCLR as needed. ZTPCLR New: send ZTPCLS over the net and then closes socket. ZTPMIO Changed to call ZTPMIR for remote tapes and FITS disk rather than ZTPMI2 and ZTPMID. ZTPMIR New: passes ZTPMIO call to network and decodes data on return. It assumes that the read buffer has been filled by the net (i.e., wait mode IO). The error code, bytes read, etc. are passed on in the FTAB. ZTPWAT Changed to call ZTPWAR for remore tapes and FITS disk files rather than ZTPWA2 and ZTPWAD. ZTPWAR New: simply returns error code from the FTAB. Moved nowhere. 6825. December 30, 1991 Special versions of Z routines Eric Changed versions for: $APLVMS/ZMOUNT Version of new ZMOUNT with no calls for remote tapes or for assigning logical AMT0n. It does check and use TAPEn. $APLSUN/ZMOUN2.C Removed all the logical name assignments (now done in ZMOUNT) and changed to AMT0n. $APLDEC/ZMOUN2.C Copy of $APLSUN one --- does not succeed in dismounting the tapes. $APLCVEX/ZMOUN2.C Removed the deassigning of the device logical and switched to AMT0n rather than MT0(n-1). $APLIBM/ZMOUN2.C Version of new $APLSUN with file locking and only a rewind (no offline) on dismount. It has to do more than the SUN one, adding a density modifier to the device name and reassigning on mount and using a ``fork'' on dismount since ``system'' does not work correctly on IBMs. $APLSUN/ZTAPE Version of new ZTAPE with an EOM operation for BAKTP and BAKLD. $APLDEC/ZTAP2.C $APLVMS version with mount removed and some error messages added. $APLCVEX/ZTAP2.C Removed ``REMOTE_TAPES''. $APLSUN/ZTAP2.C Removed ``REMOTE_TAPES''. $APLCVEX/ZTPOPN New version of $APLGEN with the quirks of file movement for Convex installed. $APLCVEX/ZTPOP2.C Removed ``REMOTE_TAPES''. $APLUNIX/ZTPCL2.C Removed ``REMOTE_TAPES'' from $APLBERK version and moved to $APLUNIX. It gives better error handling. $APLIBM/ZTPMIO Changed to call ZTPMIR for remote tapes and FITS disk rather than ZTPMI2 and ZTPMID. $APLCVEX/ZTPWA2.C Removed ``REMOTE_TAPES''. Moved nowhere. 6826. December 30, 1991 Z routines for TP networking Eric $APLGEN/ZVTPC2 Stubbed. $APLGEN/ZVTPC3 Stubbed. $APLGEN/ZVTPGC Stubbed. $APLGEN/ZVTPO2 Stubbed. $APLGEN/ZVTPO3 Stubbed. $APLGEN/ZVTPX2 Stubbed. $APLGEN/ZVTPX3 Stubbed. $APLGEN/ZVTPRC Destroy the socket from TPMON - uses ZVTPC3. $APLGEN/ZVTPRO Create the socket from TPMON - uses ZVTPO3. $APLGEN/ZVTPRX Do IO from TPMON to/from the client - uses ZVTPX3. On error, closes tape device (ZTPCLS) and resets the socket (ZVTPGC). $APLBERK/ZVTPO2.C Opens socket from client to TPMON. $APLBERK/ZVTPO3.C Create socket from TPMON. $APLBERK/ZVTPC2.C Closes sockets in client to TPMON. $APLBERK/ZVTPC3.C Destroy the socket in TPMON. $APLBERK/ZVTPX2.C Writes data to and reads from socket for the client. $APLBERK/ZVTPX3.C Reads from or writes to client for TPMON. $APLBERK/ZVTPGC.C Shutdown any connection from client and accept a new one for TPMON. Moved nowhere. 6827. December 30, 1991 UVAVG Craig/Phil Fixed the longstanding 'feature' that whatever averaging interval was inserted by the user was changed to 0.95 * that interval. On Sparc's the program would enter heavy paging mode almost immediately and would lock up the machine until it ground slowly through the averaging. Changed the ACSIZE parameter from 5,000,000 -> 2,500,000 and it now works very quickly. Moved nowhere. 6828. December 31, 1991 FRING Phil In some cases there was a problem in the lsq routine inside FRING (FRNDR) which would case the zeroth element of the Hessian matrix to be filled, causing problems elsewhere. Fixed so that if either of the addresses for this matrix are zero the routine will jump to the end of the loop. When this is in danger of happening the lsq solution never converges anyway so the solutions are blanked but it was causing funny things to happen elsewhere in the task. Moved nowhere. 6829. January 3, 1991 FRING Phil My previous change (# 6828) was not general enough. It would occasionally cause solutions not to converge in the lsq routine. Fixed it more generally so that convergence does occur but that the zeroth elements of the grad and hessian arrays are not filled in. Also changed the print level of the diagnostic messages from 1 to 3. PRTMSG will not print priority 1 messages and I have no time to find out why. Moved nowhere. 6830. 3 January, 1992 Remote tapes Eric Corrections: TPMON Raised sizes of buffers - 8197 was not enough. TPIOHD Changed to allow reading more than 1 record. IMLOD Changed to avoid a BAKF on FITS format tapes. Also corrected bug by which the same FITS disk file would be loaded NCOUNT times. PRTTP Changed to avoid BAKF's on each FITS format file. UVLOD Changed to avoid 1 of the 2 BAKFs on FITS files. ZTPWAT Changed to not report EOF from remote tapes. Moved from the VLA. 6831. 3 January, 1992 XPLOT Eric As I was recompiling AIPS, I noticed incompatible common length warnings. Someone had extra poarameters in 1 common declaration and none of the others. This caused a parameter that is used to be lost. Moved from the VLA. 6832. 4 January 1992 help files Eric Changed help files for FITLD, FITTP, IMLOD, INFILE, OUTFILE, TPHEAD, and UVLOD to explain the :: nomenclature for remote FITS disk files. Moved from the VLA. 6833. January 6, 1992 $APGNOT/FITTP.FOR Glen Caused FITTP to print the WARNING about binary files only once per execution. In a 43 file FITTP execution for athol, the AIPS message file contained over 10 pages of the same WARNING messages. Moved nowhere. 6834. January 7, 1992 ZSTAIP Eric Changed ZSTAIP to check which tapes are mounted still --- it now should know --- and to dismount them. This can really matter for remote tapes. Moved from the VLA. 6835. January 7, 1992 $APGNOT/XTRAN.FOR Glen XTRAN never worked correctly for source south of 0 dec. Program made arc seconds of dec positive when degrees were negative. Position errors of up to 60 arc seconds were estimated. New method searches for a '-' anywhere in the dec string. If one is found, the dec is assumed negative. Moved nowhere. 6836. January 8, 1992 $APLPGM/CLIP.FOR Glen Increased buffer size again to allow for many channel spectral line uvdata. Tried to remove dependence on NRPARM for determining if uv-data are compressed, instead assume data are compressed if "CATBLK(KINAX).EQ.1". Unfortunately, in some other locations where CATBLK(KINAX) is not available, if INCS.EQ.1 then the data are assumed to be compressed. Moved nowhere. 6837. January 9, 1992 UVFIT Bill C. Added adverbs STOKES and DOPOL to allow selection by Stokes' parameter and application of polarization calibration. Moved nowhere. 6838. January 13, 1992 UVFIT Bill C. Modified to give the pre and pos-fit RMS residuals in addition to the chi squares. Moved nowhere. 6839. January 13, 1992 BPASET Phil Temporarily disabled the DOBAND = 3 option, it needs work and people should not try it until it has been fixed. Moved nowhere. 6840. January 13, 1992 SNCOR Phil Added a new option to zero fringe-rates in an SN table. Also fixed a LL only bug which caused nothing to happen to LL only data. Also updated SNCOR.HLP. Moved nowhere. 6841. January 13, 1992 SNCOR Phil Needed another option, one to multiply amplitudes in the SN table. Program is nicely stuctured to be able to add new options easily. Also changed SNCOR.HLP. 6842. January 14, 1992 MK3IN Phil Several small changes: (1) Increased the number of possible CL entries from 20 per antennas per scan to 40. (2) When reading AFILE information ensured that task would only pick those type 51 extents that correspond to the type 50's listed in the IN2FILE input. (3) Also ensured that the roots and extents corresponding to the correct tape number are read, if that tape number is available. (4) Fixed a LL only problem that was causing the phase-cal information from the tape to be lost in most cases. (5) Fixed up the logic dealing with accumulation periods, they no longer have to be integral seconds. Moved nowhere. 6843. January 14, 1992 $QPGNOT/UVSUB.FOR Glen Fixed Elie Brinks spectral line LL only problem with UVSUB. Further modernized UVSUB, fixed three subroutines. $APLNOT/GETCTL assumed INCS was 3 in a funny way. $QNOT/BUFGET assumed compressed only if the vis record size was smaller after compression. For a single stokes, one channel, one IF experiment, compressed and uncompressed have the save vis record size, 8 reals. $QNOT/BUFPUT had a similar error. (I introduced the BUFGET/BUFPUT problem) Note that it was imposible to use the CATBLK(KINAX).EQ.1 test because CATBLK was not passed to the subroutine which calls BUFGET and BUFPUT. Moved nowhere. 6844. January 15, 1992 $SYSSUN/STARTPMON Pat Also $SYSDEC, $SYSIBM, $SYSCVEX. Modified this script so that it attempts to see if the TPMON daemons are already running. This script could be called from a bootup procedure but not directly as it needs the TAPEn and other environment variables defined. Moved nowhere. 6845. January 15, 1992 $SYSUNIX/AIPS.BOOT Pat Simple script to allow TPMON to be started at bootup instead of AIPS startup time. Moved nowhere. 6846. January 15, 1992 $YPGM/CNTR,GREYS $APLPGM/STARS Doug NEW FEATURE: Changed STARS to read an additional column, the position angle of the `star`. See STARS.HLP for the new file format. STARS should be backward compatible with old ST data files. The new ST files have an additional (5th) column with the position angle in degrees. Finally, updated CNTR and GREYS to read the new ST files and plot crosses rotated throughthe position angle in the ST table. Also , with STFACTOR < 0, GREYS and CNTR will add a label to each cross. The label is the row number of the cross in the ST table. CNTR and GREYS should be backwards compatible with old ST tables. Note: Other tasks that plot ST files (e.g. PCNTR and KNTR) could update but this was not done at this time. Moved nowhere. 6847. January 15, 1992 SNPLT Phil Some big changes. Modified SNPLT so that it can plot all IF dependent variables for an antenna on the same page, or the differences of the given variable between 2 specified IF's. Very useful for MkIII VLBI data. Had to add EIF to the .HLP file and also OPCODE. The user can if he/she wishes plot all the fringes rates for all IF's for all antennas in one pass of SNPLT for example. Moved nowhere. 6848. January 17, 1992 PRTAN Bill C. PRTAN now prints ellipticities and orientations in degrees as advertized rather than in radians as they are stored in the AN table. Moved nowhere. 6849. January 22, 1992 UVFIX Bill C. When UVFIX processed multiple subarrays it was successively applying the shift to the positions in each subarray. This resulted in incorrect tangent points after the first subarray and an incorrect position in the catalog header when done. The update to the position is now done only for the first subarray. Moved nowhere. 6850. January 22, 1992 GETANT Bill C. Corrected the default elipticity and orientation; the signs of the elipticity were reversed. This only affects polarization calibration capabilities not yet in the standard release. Moved nowhere. 6851. January 23, 1992 BPASS Phil Extensive modifications. BPASS was implemented in such a way that datasets with dual polzns or large numbers of IFs/channels were split up internally and the BP entries joined together in a special routine (BPUPDT). This routine was the cause of the majority of BPASS problems. Changed the size of the large arrays so that most problems will now pass through without being split up. Changed the logic inside BPUPDT so that the earlier joining together problems should no longer occur. Also made logic changes inside the main routine to tidy it up. Moved nowhere. 6852. January 24, 1992 PHSRF Ray Norris/Phil Minor fix to deal with ATAN2 blow up. Also a new ability, as well as phase-referencing, task will now perform an amplitude reference as well. Also modified PHSRF.HLP Moved nowhere. 6853. January 26, 1992 Remote tapes Eric A combination of bad logic in TAPIO and the requirement to return the requested number of bytes (no matter how many were actually read from the tape) caused the first word of the 2nd buffer to be set to zero! Changed: TPMON To copy only those words actually read and to set BUFSR accordingly in ATPMIO. ZVTPX2.C $APLBERK: To get the number of data words to read from the returned header and to return that in the call sequence. ZMOUNR To use and check a variable for number of words read. ZTAPR To use and check a variable for number of words read. ZTPCLR To use and check a variable for number of words read. ZTPOPR To use and check a variable for number of words read. ZTPMIR To set the words to write to TPMON based on FITS-standard word and to read back and copy only those data words actually returned. TAPIO To use the local word length when forcing buffer sizes. ZVTPX2 Changed precursor remarks of this stub. Moved nowhere. 6854. January 28, 1992 $APLPGM/STARS.FOR Glen/Doug W. Stars now takes 7 columns of input: The two new ones are a star type (values -1 to 19) and a 24 character label. Moved Nowhere. 6855. January 28, 1992 $YSUB/STARPL.FOR Glen/Doug W. STARPL is used by CNTR,GREYS,PCNTR and KNTR to plot stars. STARPL will now plot 21 different types of stars. < 0: No Mark, only the star label is printed 0: Cross 10: Five pointed star 1: Ellipse 11: Star of David 2: Box 12: Seven pointed star 3: Triangle 13: Eight pointed star 4: Diamond 14: Nine pointed star 5: Pentagon 15: Ten pointed star 6: Hexagon 16: 11 pointed star 7: Septagon 17: 12 pointed star 8: Octagon 18: 13 pointed star 9: Nine-gon 19: 14 pointed star >19: Ellipse The Text string labels are also ploted, unless STFACTOR is less than 0. Super-ceeds Doug W.s changes. Moved nowhere. 6855. January 31, 1992 BPASS Phil Fixed a minor logic error in the modified version of BPASS. It was not finding the correct scan if the bandpass calibrator was not the first scan in the database. Moved nowhere. 6856. January 31, 1992 HF tables Bill C. Two new routines, HFINI and TABHF were added to APLNOT. There routines create/initialize and do I/O to HF tables. These tables are written by MK3IN to contain Haystack FRNGE (hence HF) results. These tables are straight dumps of the A tape 4000-4500 records. A description is given in the MK3IN.HLP EXPLAIN section. Moved nowhere. 6857. January 31,1992 MK3IN Bill C. Modified to read the type 4000-4500 records (FRNGE results) and dump them to an AIPS table of type HF. A detailed explanation of the contents as revealed from Haystack documentation is given in the EXPLAIN section of MK3IN.HLP. Moved nowhere. 6858. February 3, 1992 CHNDAT Bill C. This routine was using the axis number of the frequency axis in the catalog header from the standard (UVPGET) common. This may give incorrect results especially when used in the calibration package which may reorder the axes. CHNDAT now searches for the "FREQ" axis. The result is that the channel increment returned by CHNDAT may be VERY wrong. Moved nowhere 6859. February 4, 1992 MK3IN Bill C. The sign of the phase of the phase cals were wrong. The sign of the data phases were corrected some time ago but the phase cal phase was not. This correction should cause the phase cals to (nearly) line up the IF phases again. Moved nowhere. 6860. February 4, 1992 POPSDAT.HLP Dave A. Changed the default value of SUBARRAY from 1 to 0 so the default will include all antennae. Moved nowhere. 6861. February 5, 1992 $APLGEN/ZTXOPN,ZTXOP2 $APLCVEX/ZTXOP2 Glen Added two new opcodes for Text file opens, 'QRED', and 'QWRT' which are quiet read open and quiet write open. This change was made to reduce confusing messages when starting up AIPS in the Network Configuration. Moved Nowhere. 6862. February 6, 1992 CLCOR.HLP Bill C. Corrected the table of vapor pressure of water; the old values were a factor of 10 too high. Moved nowhere. 6863. February 6, 1992 MK3IN Bill C. Changed the default number of lags to 8, the normal continuum mode. Under some circumstances this was never getting set and a value of 0 causes MK3IN to selfdestruct in obscure ways. Also MK3IN now traps some invalid values of the number of polarizations, frequencies or IFs and attempts to give meaningful messages and dies. Also removed the dire warnings about using the program - there is now a reasonable expectation that it will work. In the EXPLAIN section section of the MK3IN.HLP file there is now a discussion of how to deal with multiband (e.g. S/X) observing. Due to the use of the "frequency codes" by the Haystack system to denote either the polarization product or the frequency band only one band can be processed ata a time. Moved nowhere. 6864. February 18, 1992 SNCOR Bill C. Added options 'CLPD' and 'CLPR' to flag solutions with delays or rates outside of a given range. Also SNCOR.HLP. Moved nowhere. 6865. February 18, 1992 PHSRF Phil Made the modification sent by Ray Norris (#6852) more general. Will now write the full complex reference to a BL table as well as correct the data with it. Changed one of the OPCODE's and added another in the process. Moved nowhere. 6866. February 18, 1992 CGASET Bill C. Reduced the time between interpolations of the CL/SN table entries from 20% to 10% of the interval between the entries. Moved nowhere. 6867. February 19, 1992 PCAL,PCAL.HLP Bill C. Made serious changes/additions to the way polarization calibration can be done to allow the calibration of VLBI data. The option 'APPR' will only work properly for arrays with antennas which all have the same parallactic angle (OK for VLA but NOT VLBA). The new options will support partially resolved calibrators using the assumption that the Q and U images are scaled versions of the I image. In general this is not very good but is OK for unpolarized sources. Both of the new solution methods require that the parallactic angle have been removed from the data prior to phase calibration. This is the opposite of the PCAL 'APPR' solution. Added a solution option (RAPR) for fitting a linearized polarization model allowing some source resolution. This is implemented in the new routine RPCALC. This routines uses a model suitable for VLBI polarization data and the solution type in the AN table is given as 'VLBI'. Implemented solution option 'ORI-' to fit ellipticities and orientations. This is needed to model poorly matched sets of feeds. This involves solution routine IPCALC and related routines. This option adds a whole load of user options. This method allows a partially resolved source. This method takes rather alot of computer time to run as it is a seriously nonlinear model. Also fixed a bug (feature?) in PCLAVG which in combination with other features caused partially flagged data to result in completely trashed averaged data. Also PCAL.HLP. ALso fixed logic problems when averaging in IF; only the first IF was being used. Also fixed numerous bugs which caused PCAL not to be able to use a provided polarization model. Also fixed several problems with multiple IFs, especially when an averaged solution was obtained. Moved nowhere. 6868. February 19, 1992 DATPOL, POLSET, VLBPOL,ORIPOL Bill C. Modified the polarization calibration routines to handle solutions from PCAL options 'RAPR' and 'ORI'. Added routines VLBPOL and ORIPOL. Moved nowhere. 6869. February 19, 1992 PARANG Bill C. The old version computed the parallactic with a one argument atan function. This has quadrant ambiguities. The correct method is to use a 2 argument atan function. Moved nowhere. 6870. February 19, 1992 New polarization poln. parameters Bill C. The 'ORI-' solutions from PCAL require two additional parameters which are stored in the AN table. These are the R-L phase difference ('P_DIFF') and the polarization reference antenna ('P_REFANT'). New routines PDRSET and PDRGET write and read these values from the AN table. DSEL.INC is modified to carry these values. In DSEL.INC also moved the big buffers to a single common to cure some alignment problems with equivalences to buffers. Moved nowhere. 6871. February 24, 1992 ANCAL Phil A temporary array into which KEYIN fed the CAL file was not big enough for some VLA cal files. Increased its size. Moved nowhere. 6872. February 25, 1992 SN2CL Bill C. Under some circumstances calibrators which were not selected were being used anyway. The routine now blanks solutions for non selected SN table entries to insure that they are not used. Also beefed up the logic to keep from writting a final totally blanked entry at the end of the table. Moved nowhere. 6873. February 25, 1992 $AIPSUB/KWICK Glen KWICK was causing a segmentation violation when the arguments to an AIPS procedure did not match those expected by the procedure. Added an error check, but has a fixed number for the CORE size. (It is not parameterized in COMMON) Moved Nowhere. 6874. February 25, 1992 $SYSUNIX/OPTNOT.LIS Glen List of subroutines which should not be optimized by the compiler. No Changes to COMRPL have yet been made. Moved Nowhere. 6875. February 25, 1992 $SYSUNIX/WHICH Glen Little script to Find aips software within an AIPS version ie % WHICH MX yields WHICH MX*.* file(s) are below /aips1/code/15APR92 /aips1/code/15APR92/HELP/MX.HLP /aips1/code/15APR92/QY/PGM/NOTST/MX.FOR Moved Nowhere. 6876. February 26, 1992 $APLBERK/ZABORS.C Chris, Pat Actually it was a gathering of AIPS that found this little problem. It appears that a floating point exception caused the exception handler to kick in. It in turn called ZMSGWR which must have caused another exception and so ZABORS got called recursively. This may be the key to the convex (and other systems) lock file problem. Chris put in code that prevents this race condition from developing. This was on the convex. Other areas affected are APLIBM, APLDEC, APLSUN, APL3090 and APLUNIX. Moved Nowhere. 6877. March 3, 1992 Aborts Eric Changed $APLGEN/ZSTAIP.FOR to test that its task name is AIPS, GR2TEX, or GRITP before doing tape dismounts. Changed ZABORS.C in $APLSUN, $APLIBM, $APLDEC, $APLBERK, and $APLUNIX to call ZSTAIP. Moved nowhere. 6878. March 3, 1992 POPS Memory files Eric Changed the POPS memory files to have more space: 950 -> 1450 words of temporary literal storage 14760 -> ..... words of program storage 4173 -> .... words of variable storage 4096 -> 10240 words of source (text) storage In addition, the text storage is now kept in core rather than in a scratch area in the MEmory file. MEmory files are also changed to have a RESTORE 0 file in a release-dependent area and a RESTORE 1 file in each of the machine-dependent $DA00 areas. Changed: DAPL.INC Raise K array size and add LISTF, STORE1 and STORE2. DCON.INC Raise K array size and add LISTF, STORE1 and STORE2.. AIPS Changed initialization of temporary literal structure. Init STORE1. AIPSB Changed initialization of temporary literal structure. Init STORE1. AIPSC Changed initialization of temporary literal structure. Init STORE1. FILAIP Changed ME file size (377 -> 150) and made it created 0-n rather than 1-n. FILAI2 Changed ME file size (377 -> 150) and made it created 0-n rather than 1-n. POPSGN Change to write only the basic, version-dependent, but not POPS-number-dependent, ME file. Drop write/read to LISTF scratch disk area. Change various size parameters. INIT Init STORE2 and use new values for K array sizes. Drop write LISTF to disk. STORES Drop I/O to scratch area of disk for LISTF. Test STORE1 to determine if any recent STORE 1 was done and STORE2 to see if someone else did it. Changed addressing to disk as well --- correcting a 1-block error from before. ZPHFIL ($APLGEN) Changed to reference version-specific ME files only for sequence 0. STORE.HLP Changed wording. RESTORE.HLP Changed wording. AU2A Changed SG version number to 13. SGLOCA Changed SG version number to 13. KWICK Changed bad address tests to use variables from common. Moved nowhere. 6879. March 5, 1992 $SYSAIPS/AREAS.DAT Pat Removed the following (unused) areas: YSSS, YSVU, YVDEV, YXTV, YXTVUNIX, YXTVBELL, YXTVBERK, YXTVVMS. Moved Nowhere. 6880. March 5, 1992 $SYSUNIX/* Pat, Mark Implemented a scheme similar to that described in AIPS memo 74 (the AIPS setup at ATNF). Significant directory changes include: creation of $ARCH (architecture) directories under the $AIPS_VERSION area, moving $SYSLOCAL (and the SYSIBM, SYSDEC, SYSSUN, etc areas) to $AIPS_VERSION/$ARCH/SYSTEM, adding the $PREP area, moving $LOAD and $LIBR and $MEMORY under $ARCH also. New scripts include AIPSASSN.CSH and .SH, AIPSPATH.CSH and .SH, AIPSROOT.DEFINE, HOSTS.CSH and .SH, PRDEVS.SH, START_AIPS, START_SERVERS, TVALT, and TVDEVS.SH. Also SYSETUP, DASETUP. Almost all of the other $SYSUNIX scripts were modified, some trivially and some extensively. The AIPS script is gone and is replaced by START_AIPS (run via "aips" and "AIPS" symbolic links, usually from /usr/local/bin or /local/bin). New setup files include NETSP and HOSTS.LIST (plain text). Printing scripts ZLASCL and ZLPCL2 are now generic and should not have to be edited at all (PRDEVS.SH is edited instead). TEKSERVER has been added. There's more, and it will be covered in an upcoming AIPS memo (soon, I hope.) Moved Nowhere. 6881. March 6, 1991 SPLIT Neil Killeen/Bill C. In SPLTUV DOBAND was being saved in a temporary to keep UVGET from initializing the bandpass calibration when UVGET was only being called to create the output catalog header. DOBAND was not being reset untill after the error check for UVGET and so if data for a source was not found DOBAND was not being reset to the original value. Moved DOBAND = SAVBND to before the error check. Moved nowhere. 6882. March 6, 1992 MK3IN.HLP Bill C. Added documentation describing how to deal with DAT tapes from the correlator containing copies of multiple archive tapes separated by EOI marks. Separate runs of MK3IN are required to read in the separate tapes and some systems may refuse. Moved nowhere. 6883. March 8, 1992 $INSUNIX/INSTEP2 Pat Modified this script so that it creates the MEMORY area in the right place ($AIPS_VERSION/$ARCH/MEMORY). Also modified its check on the ALT areas so that it keys off the ":n:" fields in the $SYSLOCAL/LIBR.DAT file instead of whatever AREAS.SH/CSH define. It also defines LOADn if they need defined and alerts the installer that they were undefined. Moved nowhere. 6884. March 9, 1992 $SYSUNIX/AREAS(C)SH.SED Pat These scripts were mis-defining MEMORY. This area is now only dependent on the architecture since Eric split the memory file in two. Hence MEMORY is $AIPS_VERSION/$ARCH/MEMORY. Previously it was $AIPS_VERSION/MEMORY/$HOST. The host- specific part of the memory files now resides in $DA00, one for each POPS# as before, while the MEMORY area now only contains one file. Moved nowhere. 6885. March 10, 1992 UVGRTB Bill C. This routine was using the first frequency in the data rather that the highest frequency being gridded to reject data outside of the grid. This caused problems when the data was heavily tapered and uv data lay outside of the uv grid. Random errors and aborts could result. Moved nowhere. 6886. March 11, 1992 HORUS Bill C. Routine MAPRIN was making the output files for OPTYPE='SUM' images too large. The output file was too large by the number of channels/IFs averaged. Moved nowhere. 6887. March 11, 1992 UVGRTB Bill C. Modified logic to reject data within a half width of the convolving function kernel of the exterior edge of the grid in both u and v. Formerly the logic allowed the convolving kernal to wrap around from one edge in u to the other. Moved nowhere. 6888. March 12, 1992 STORES Eric It was using LISTF as a scratch array which, since it now in COMMON, is a naughty thing to do. Especially in this case where a scratch array was available and the use was in the read step when it was advertised to be in the write step. Corrected that too. Moved nowhere. 6889. March 12, 1992 $APGNOT/HOLGR.FOR Mark New task to process antenna holography data. Uses the following routines which were also checked in: $APLNOT/FFT.FOR $APLNOT/FFTV2.FOR $INC/FFTV2.INC $INCVMS/FFTV2.INC $APLNOT/LEASQR.FOR Moved nowhere. 6890. March 12, 1992 ZIVSOP and FILINI Pat Fixed ZIVSOP (Ivas open routine) so that it allows for 255 TV's. Also fixed FILINI so that it does Image Catalog initialization properly (it had not been setting the TV# in a loop over all TV's). Moved nowhere. 6891. March 13, 1992 HORUS.FOR Bill C. No longer modifies frequency alternate axis reference pixel. This is now done in IMCREA. Also added more calibration information to the history file. Moved nowhere. 6892. March 13, 1992 IMCREA.FOR Bill C. The reference pixel was being incorrectly set when channels were being averaged. Now if channels are being averaged the reference pixel is set to 1.0 and the correct frequency is assumed to be in the common variable FREQ. If there is no averaging the old values are kept. Moved nowhere. 6893. March 13, 1992 PRTUV.FOR Bill C. Modified to give a message about the printed u,v and w being in wavelengths at the reference frequency. Moved nowhere. 6894. March 13, 1992 SHOUV.FOR Bill C. An uninitialized variable was causing a zero divide in SPECUV. Moved nowhere. 6895. March 14, 1992 UVCOP Bill C. / Neil If data were selected by FQ ID then the u,v and w were not scaled to the new frequency unless selection by frequency or IF. Also, UVCOP was mucking up the FQ table adjustment when the FREQID selected was greater than zero. This bug was inserted in the recent CHNDAT change. 6896. March 16, 1992 FRING Bill C. In the least squares routines data from antennas flagged in the FFT search was not being ignored. The mannar in which the data was being used depended on recent successful fits. The most common symptom was the entire solution failing. The problem was most serious for low signal-to-noise cases. 6897. March 18, 1992 MX-$APLSUB/METSCA.FOR Glen MX was getting a segmentation violation in METSCA when large values were passed to METSCA. Added array bounds checks. Now NaNs are printed. Moved Nowhere. 6898. March 18, 1992 UVSUB-$APLNOT/BUFXPN,BUFPAK Glen Eli found problem for 2 stokes data. Since compressed UV data are expanded in place, moving the random parameters can over write the random parameter in the same visibility for a small number of correlators. Modifed pair of routines to copy random parameters to a buffer before expansion/packing. Moved nowhere. 6899. March 18, 1992 $APGNOT/PRTUV.FOR Glen PRTUV would not print bad times because of time selection defaults. Modified PRTUV to print bad times if no time selection is specified. Moved nowhere. 6900. March 19, 1992 $APGNOT/FILLM.FOR Gareth Three trivial mods: 1) If the recorded integration time was zero (as was the case on some ancient archive tapes), assume 10 sec. 2) Weighting is now more precisely done: 10 sec. ==> 1.00000 (was 0.99840) 3) The pseudo-(u,v,w) values stored in Holography mode are now as advertized. Whether this is what we want, however, ... Moved nowhere. 6901. March 20, 1992 AFILE Phil/Neil Neil found a lot of the comment lines describing the A-file format were > 72 chars, corrected. Moved nowhere. 6902. March 20, 1992 VLBIN Phil After much pain Tony Beasley found that the sideband recorded on MkII correlator tapes is not necessarily the sideband observed by the VLBA. Added an option (VLBINPRM(16) - this is ridiculous) to enable the user to set the true sideband of the data. Most of the time setting this to zero will do the right thing. Moved nowhere. 6903. March 20, 1992 SNPLT Phil Added two new features: (1) When plotting tables if a value is blanked it will now be plotted at value zero with a different symbol from all other values. (2) Enabled SNPLT to also plot the SNR (or weight) column. Also modified SNPLT.HLP. Moved nowhere. 6904. March 20, 1992 $SYSSUN/COMLNK Pat This only affects the NRAOCV site. Changed the logic that was putting the records in the transaction files for the midnight job. Previously it tried to add anything (a) without checking if the file was in the $AIPS_VERSION directory tree, and (b) before SEARCH looked for the file. The result was a lot of bad records in the .UPD transaction files. Now it checks these two things before updating the transaction file. So there should be no more false entries in the COMLNK.UPD file (I hope). Moved nowhere. 6905. March 20, 1992 $APLSUB/GALPOL.FOR Glen Created a new subroutine which will return the ra and dec of the North Galactic Pole as a function of Epoch. Currently only 1950 and 2000 (the default) have values. This subroutine is used by $AIPSUB/AU7.FOR and $AIPSUB/COORDT.FOR to translate from RA and Dec to Galactic Coords. Must be updated for 2050. Moved Nowhere. 6906. March 20, 1992 GETHUT, FITTP Phil There was an assumption inside GETHUT and FITTP that there could be no tables with > 50 columns, this violates FITS standards as well as screwing up when encountering the new HF tables which have 77 columns. Fixed the array declerations in both these routines. Moved nowhere. 6907. March 23, 1992 TVFLG, SPFLG Eric Changed GRIDTB to use a single subarray when specified. It was trying to use SUBARRAY through max-subarray in building the grid file. Also changed SPFOAD and TVFOAD to have a friendlier error message when the integration is too short in computing RMS. Also changed the handling of errors in the main function routines of SPFLG and TVFLG. Error conditions were getting masked with no output message. Moved nowhere. 6908. March 24, 1992 DATPOL.FOR Bill C. This routine was not properly addressing visibillity data if the Stokes axis was other that the second regular axis. Moved nowhere. 6909. March 24, 1992 FRING Bill C. In the FFT search the "AP" memory was not being cleared before use. This caused hystersis effects for MkIII data when there were variable number of time samples in a SI (usually the case). The effects were usually spurious false detections on weak fringes. If there is insufficient memory for at least a factor of 2 over sampling in the FFT search then an informative message is printed and the task quits. Also made some cosmetic changes with involving the print of the results. Moved nowhere. 6910. March 25, 1992 $APGNOT/FILLM.FOR Gareth I believe I have (at last!) found the bug which caused the date to be incorrect for the first visibilites of a FILLM run when appending to an old file. I have also changed the way that the VLAOBS parameter works when set to blank. Previously, FILLM filled all data matching only the VLA observing program of the first data record matching the other selection criteria. Now, FILLM will fill ALL data matching the other criteria, i.e., blank means select ALL data. Moved nowhere. 6911. March 25, 1992 CVEL Phil Major rewrite. CVEL is one of the few (the only?) task that can read a multi-source file and has to write a multi-source file changing things on the way. The way this was done was via UVGET, this had unpleasent consequences for the CATBLK of the o/p file, and required much futzing around with the header. Rewrote the task to use UVINIT, UVDISK - this simplifies the logic enormously and should make program maintanence much easier. One nuisance involved in this is thet CVEL now has to deal with the NX table itself and also to do its own BP correction, subroutines were added to do this. Also modified the messages the user receives along the way. Moved nowhere. 6912. March 25, 1992 DBCON Phil/Bill In the case of a single source file DBCON was not setting the source number correctly. It was accessing an incorrect value inside the random parameters array and passing IROUND a NaN. Moved nowhere. 6913. March 25, 1992 FITTP Phil The translation of the FQ -> CH tables has caused much confusion over the past couple of years, and also sometimes has a problem in that the translated CH table does not get written to tape. We feel that enough time has now passed since the FQ tables were invented (glorious day!) that they should now be the default. If a user really wants to take a CH table away (assuming that is possible i.e. only 1 FREQID) then they have to set DONEWTAB to 3. This will write all tables as FITS binaries but will do the translation from FQ -> CH. Also updated FITTP.HLP. Moved nowhere. 6914. March 26, 1992 MK3IN Neil/Bill C. Removed tabs from FORTRAN file. MK3IN now makes an AN table with all antennas specified by the user including any 'ANY' antennas. Changed parameter MAXM3X to 50 to handle FRNGE output for up to 50 baselines. Modified MK3IN to read a flag in the experiment (INFILE) file which tells if multiple polarizations are to be read. If not, multiple frequency codes can be read (e.g. S/X) into a single AIPS data file which then must be seperated using UVCOP. The default polarization is now LL (the standard MkIII astronomical setup). Also modified the description in MK3IN.HLP. Moved nowhere. 6915. March 26, 1992 SNCOR Bill C. Added option 'REFP' to reference all the phases to a single IF. This makes it simpler to look for IF-IF phase differences. Also SNCOR.HLP. Moved nowhere. 6916. March 26, 1992 FRING Bill C. Fixed error in previous fix to FFT fring search. Moved nowhere. 6917. March 27, 1992 $RUNSYS/VLACLOAD.001 Glen Small bug in VLACLOAD was not allowing the test disk (TDISK) to be on a different disk than MDISK for test 32. (The SPLIT test). Modified OUTZAP to delete all files matching the output name on all disks, ie INDISK=0, not INDISK=TDISK. Moved nowhere. 6918. March 27, 1992 $APGNOT/CLCOR.FOR Glen Changed the meaning off the opcode 'PHAS' and added opcode 'RATE' which does what 'PHAS' used to do. 'PHAS' now adds a constant term to the orientation angle for 1 or more IFs. With 'PHAS', there is no time varying term and there is no modification of the AN table. The opcode 'RATE' modifies all IFs specified with BIF and EIF in the same way. Also added Bill C's modification to CLCOR.HLP. Moved nowhere. 6919. March 29, 1992 AOC AIPS installation Mark Fixed bugs in the following startup scripts - the main problem was with hard-wired TVs and starting the TPMON servers AIPSEXEC AIPSPATH.[C]SH START_AIPS START_TPSERVERS START_TVSERVERS STARTPMON TVDEVS.SH Removed START_SERVERS, its function is now split between START_TPSERVERS and START_TVSERVERS. Also, two new utility scripts ALLRSH DATOUCH Fixed the include path for SUNs in $SYSLOCAL/INCS.SH so that it searches $INCSUN, and added a SUN-specific version of DAPC.INC in this directory (all Q-routines and programs for SUN4 should be recompiled). 6920. March 30, 1992 APLNOT/DATCAL Phil Minor modification to ensure that the (computationally) expensive delay and delay-rate corrections are not made to the data unless they have to be. Moved nowhere. 6921. March 31, 1992 $QYPGNOT/MX Glen 'Minor' modifications to allow larger section of the Beam to be used in the minor cycle and also fewer residuals. Currently the maximum number of residuals loaded into the AP is 3000. This improves cleaning of badly calibrated data by allowing fewer side-lobes to be selected as source components. Moved nowhere. 6922. March 31, 1992 IVAS Eric Changed $YIVAS versions of YCUCOR Fixed major errors! It was using a scroll on input that was not set and should not have been used. Then it used the wrong variable name in computing the output locations leading to error of several pixels. YSCROL Changed to use separate FIVASVPZOOMSCROLL calls for grey and graphics, and for CV only at present, to do a minor correction (0,1,2 pixels) in the graphics. YZOOMC Changed as for YSCROL. YLUT Changed to do gamma correction when the IVAS is in the 3-channel mode (where LUT is OFM). YOFM Changed it to avoid negative OFMs which arose on read. YCHRW Changed it to be similar to the generic routine for lettering in the grey planes. It did not handle non-standard sizes well. Moved nowhere. 6923. March 31, 1992 TVs Eric In messing with the IVAS, made some other changes: GRBOXS Changed to do simple rounding rather than down at BLC and up at TRC. IMANOT Changed call sequence to add the desired graphics channel of the lettering. IAXIS1 Changed for call sequence of IMANOT. ITICS Changed for call sequence of IMANOT. AU5B Changed for call sequence of IMANOT. AU5E Changed to have graphics controls in lower-level code. OFMCON Changed to use IMANOT to make lettering more legible (against a black background). OFMGAM Changed to use IMANOT to make lettering more legible (against a black background). OFMMOD Changed to use IMANOT to make lettering more legible (against a black background). IMCCLR Adjusted brightness of levels 1, 2, and 5 of basic color contour set. Moved nowhere. 6924. March 31, 1992 TVANOT Eric Changed logic of how it chooses the Graphics or Grey plane to letter. This will atke advantage of the improved TVANOT and be more flexible. Changed AU5B.FOR and TVANOT.HLP. Moved nowhere. 6925. April 1, 1992 FILINI Eric The loop that initialized TVs had several errors: it told the current TV number with an uninitialized variable, it skipped an introductory message, and it used uninitialized common variables NGRAPH and NGRAY. Reordered the initialize call to YTVCIN, added MSGWRT call and a second message, and added the ability to skip out of the loop entirely. Moved nowhere. 6926. April 2, 1991 FRING Bill C. Fixed indexing problems in the least squares routines that caused a bad display of the fitting results. The fits themselves were unaffected, only the optional display of the results. Also modified the least squares routines to include the sum of the weights in the penalty terms for the delay and rate window. Moved nowhere. 6927. April 2, 1992 New task: SBCOR Phil A new task. SBCOR will correct VLBI data that contain mixed MKIII/VLBA baselines for the 130 degree phase offset that occurs between upper and lower sidebands. Hopefully this task will be short-lived and we insert this correction into MK3IN. Also a helpfile. Moved nowhere. 6928. April 2, 1992 POLSET Bill C. In the call to PDRSET the value of NUMIF in common was assumed to be set. Under some circumstances it was not. EIF replaced the call to PDRSET as this should always work. Moved nowhere. 6929. April 3, 1992 ORIPOL Bill C. Modified the Stokes transformation matrix to rotate RL and LR data by the R-L phase differences given in the antenna table. This appears to allow CLCOR to rotate the polarization angle. The interaction of this use of the R-L phase difference and that in PCAL is uncertain. This affects the orientation-ellipticity model ('ORI-' in PCAL) only. Moved nowhere. 6930. April 3, 1992 CLCOR Bill C. Modified 'POLR' option to deal with ORI-ELI poln models by modifying the R-L phase differences in the AN table rather than changing the polarization parameters. Moved nowhere. 6931. April 3, 1992 LISTR Phil Several small changes. The most significant being that the problem with the phase averages after matrix listings being incorrect when the points lay near 180 degrees is now fixed. When listing gain entries from tables with > 1 IF, the frequency labels for the IF's were wrong, they were always the label for the first IF. A cosmetic change suggested by Glen for deciding when to start a new page. The qualifiers are now modulo 10000 before being printed in order to deal with large numbered qualifiers. A change suggested by Gareth. Moved nowhere. Its somebody else's baby now. 6932. April 6, 1992 PRTAN Bill C. Modified to print the Right-Left phase differences and the polarization calibration reference antenna for the ellipticity-orientation polarization feed model. Moved nowhere. 6933. April 6, 1992 $QYPGNOT/MX MX.HLP Glen Backed out of some of the Minor Cycle changes put in on March 31, 1992 inorder to return MX to its former speed. Have removed the MINPATCH upper limit, so that MINPATCH up to 179 is allowed (Beam width of 359 pixels max). This helps when doing snap-shot imaging. Initial clean step use more of the beam, but as clean reaches noise limit, MX returns to its former method. Updated MX.HLP. Moved nowhere. 6934. April