----------------------------------------------------------------------- ; Copyright (C) 1998-1999 ; 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 Master version of AIPS kept in Charlottesville. An entry is: line 1: Date Program name Your name line 2: What was done (may span multiple lines) line n-1: Where has this change been moved (e.g. 15JUL94, nowhere) line n: 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 15OCT98 *********************************************************************** *********************************************************************** Changes while 15OCT98 is TST *********************************************************************** *********************************************************************** 9810. March 19, 1998 24-bit TV Eric Rewrote parts of XAS to support 24-bit TrueColor TVs with an IIS Model 70 like algorithm when the user requests it (.Xdefaults parameter) and the display host allows it. Other changes: YSPLIT To send all data to XAS under new opcode 47. SSS.TXT To document YSPLIT change TVMOVI Added frame rate testing if POPS debug is on. ZSSSO2.C Added SPLOT opcode 46, SPLIT -> 47. XAS.HLP To document all of the CVX changes plus this change. To use YHOLD a good bit more to avoid excess blits in this mode (n.b. a screen re-write is needed whenever channel or graphics selection changes, whenever LUT of OFM changes, zoom changes, etc.) If several of these are to be done, issue a YHOLD onnn before them and a YHOLD offf after them. $AIPSUB/AU5, AU5A, AU5B, AU5D, AU5E, AU6, AU6A, GRPOLY, GRLUTS, HILUT, TVMOVI, TVROAM $YSS/YLUT $YSUB/COLORC, OFMIO, COLORL, COLORF, GRBOXS, IENHNS, TVBLNK Moved many of these to 15APR98 this date. 9811. March 19, 1998 INDXR, CLCAL Eric Blocked the VLB mode in INDXR and dropped the subarray 0 output option in SN2CL and CLCAL. They cannot work as intended given the rest of the cal computation and application code. A special task to make a regularly gridded (properly chosen) version of the multiple subarray CL table will need to be written to achieve the desired results. Moved to 15APR98 this date. 9812. March 19, 1998 $APLNOT/MCTAB Ketan Simple typos caused UVCOP to crash, two references to RECR should obviously have been to RECD. Recompiled UVCOP, anything else? Moved to 15APR98 this date. 9813. March 19, 1998 $APGNOT/MSORT, $HLPFIL/MSORT Ketan Yet another UVDISK problem related to sorting very small files. Also added a PRTLEV to make output more useful. Moved nowhere. 9814. March 19, 1998 $QPGNOT/KRING Ketan Simple typo resulted in multi-band delays being 10^9 too large. Moved to 15APR98 this date. 9815. March 23, 1998 $APLNOT/INDXIN Ketan INDXIN did not properly check the FREQID and NXSUBA values coming from the NX table or the SUBARR value specified by the calling task. This could sometimes result in an NX record being delivered that shouldnt be. This could only have affected VLBI data and only when actually subarraying sources. Moved to 15APR98 this date. 9816. March 26, 1998 $INSUNIX/CDSETUP, INSTEP1 Pat New script to do the bridging work from a CDrom to the start of INSTEP1. It will permit the installer to (a) run from the CD [with a mere 10 MByte footprint on hard disk], (b) install a full-blown system, or (c) switch from one to the other. Added hooks in INSTEP1 so that it doesn't try to move files off read-only media or equally silly things. Moved to 15APR98 this date. 9817. March 30, 1998 $SYSUNIX/XASERVERS Pat Changed the order in which daemons are started: do the TVSERV and XAS servers first. Robert Reid at Toronto found this to avoid a problem when using ssh to start up the servers remotely. Also removed the resetting of the DISPLAY to :n from hostname:n (this really messes up ssh connections with X tunneling) and streamlined it a bit (check on LOAD area only needed once). Moved to 15APR98 this date. 9818. March 30, 1998 Misc Eric Corrected minor typos in Chapters 1, 4, and 5 of the CookBook. Cleaned up the messages (old debug junk was left) in XAS.SHR. Moved nowhere. 9819. April 1, 1998 XAS, etc Eric Changed: XAS.SHR Fixed coloring of graphics planes, byte order for PCs, zeroing of individual graphics channels TVRGB Moved and added YHOLDs, zero LUT(1), OFM(1). TVHUI Raised internal buffer size. Moved nowhere. 9820. April 2, 1998 $APGNOT/ACCOR.FOR Leonia ACCOR has not worked if the selected data have many negative weights. In this case ACCOR continues accumulating points with both negative and positive weights. That is not correct because points with negative weights should be rejected. I modified ACCOR to reject the points with negative weights and change some condition to exclude completely possibility of under SQRT expression to be negative. Moved nowhere. 9821. April 2, 1998 $QPGNOT/FRING.FOR Ketan The DOFIT option in FRING did not work correctly if APARM(9).LE.0 due to an undefined indexing array. Now fixed. Moved nowhere. 9822. April 2, 1998 $QPGNOT/KRING.FOR Ketan The SNR calculation for the FFT stage of KRING has now been retro-fitted to match Bill C.'s original formula. This is so people can are presented with a formula whose SNR calculations may be wrong but, at least, are familiar. .HLP file updated. Moved nowhere. 9823. April 2, 1998 $RUNSYS/REFREQ.001 Ketan REFREQ will now no longer allow the user to shift reference frequency because this was not implemented consistently [the UVW coordinates were not properly adjusted]; implementing this would require REFREQ to become a task and it is not clear that the GEODESY community is well served by this capability anyway. .HLP file updated. Moved nowhere. 9824. April 2, 1998 $APGNOT/UVPRT.FOR Ketan If SUBARRAY=0 is set in UVPRT, the same visibility can get printed to the user multiple times. The potential confusion outweighs the inconvenience of only being able to view one subarray at a time. .HLP file updated. Moved nowhere. 9825. April 3, 1998 $SYSUNIX/OPTIMIZE.LIS Eric/Pat Added an entry for DECBIT (in $APLSUB) to force no optimisation. The OPT3 level apparently did a number on this subroutine, causing bizarre behaviour such as TVLOD doing next to nothing other than saying "TV PROBLEM". (The channel to load was getting passed to TVLOAD as zero). Setting the level to OPT0 for just this one routine seems to have cured the problem. Moved to ohsumi this date; will move to 15APR98 shortly. 9826. April 4, 1998 PRTTP,PRTLIN,TXPL,AU1A,CATLST,PRTALN Jeff All references to CRTMAX changed to ABS(CRTMAX)...prepping for auto-dectection of window-size changes. Moved nowhere. 9827. April 5, 1998 STARPL Eric Strangely the integer star type is a float in the table. So a float got passed to PNTPLT which expects an inteeger. Fixed up STARPL to use an integer, reading a float from the table. Moved to 15APR98 this date. 9828. April 9, 1998 $APLNOT/GETTIM.FOR, CVEL.HLP Leonia CVEL calls the task GETTIM.FOR which looks for the array name. If aparm(7) = 1, it expects the VLA/ATCA/LBA. Data from the Australian S2 correlator is labelled 'ATLBA', not 'LBA' so GETTIM fails (and so does CVEL). GETTIM should check for both 'LBA' or 'ATLBA'. I added ATLBA to the list of 'VLA' group of arrays (in GETTIM). A small bug is fixed at GETTIM as well: If APARM(7) = 0, expecting VLBA, GETTIM failed if data are from VLA, but it shoud fail for any array belonged to the 'VLA' group: VLA/ATCA/LBA/ATLBA. The relevant change of CVEL.HLP has been done also. Moved nowhere. 9829. April 9, 1998 XAS.SHR Eric Changed code to do x and or y split screen. This enables roam and other algorithms. In the 8-bit PseudoColor mode, only one lookup table can be active. This means that the option in blink that allows either LUT to be changed while looking at a split screen only shows the changes to one of the two images. The other does get changed, but you see that only when the blink resumes. It works very well indeed in the 24-bit TrueColor mode. Changed TrueColor to be the default. On a computer that does not do TrueColor there will be a message about goping to PseudoColor. This message may be suppressed by a AIPStv*useTrueColor: 0 in your .Xdefaults file. Moved nowhere. 9830. April 10, 1998 IMAGR Eric Fixed it to be more understanding about a missing AN file. It can't get the average parallactic and zenith angles, but usually one does not care. Moved nowhere. 9831. April 13, 1998 RESEQ Chris Fixed several gross errors. - ZUVXPN and ZUVPAK are now called for compressed uv files and not for uncompressed files (this was reversed in previous versions). - Cross polarizations are now exchanged when antennae are switched on a baseline. - The baseline vector is now multiplied by -1 when antennae are exchanged on a baseline as required for consistency. - UV buffers are now sized using UVBFSL from ZPBUFSZ.INC. - UVINIT is given the correct buffer size in GBLOCK thus avoiding erroneous complaints about the buffer being too small. Moved nowhere. 9832. April 13, 1998 CXCLN Chris Attribute dimensions are now set correctly when storing the default window to the INPUTS object allowing CXCLN to run without specifying an explicit window. Moved nowhere. 9833. April 13, 1998 CCEDT Chris Added a crude hack to avoid division by zero in the loop that subdivides the region being searched for model components into rectangular subregions. Moved nowhere. 9834. April 14, 1998 OOP Eric Changed OBJMNGR to limit LUNs to 120. For some reason ZOPEN, ZFIO, ZMIO have this limit rather than 128. Corrected QCLEAN - the NDIM argument to ARRWIN needs to be dimensioned (7) to hold all axes. It was a scalar causing the HP to go into an infinite loop. Moved nowhere. 9835. April 16, 1998 CCMOD Eric This old clunker never worked. Thanks to Lincoln Greenhill for pointing out one of the major bugs. Fixed: 1. Error/file status handling 2. Addressing in the array (left and lower rows were omitted) 3. Computation of disk - the radius was never tested so it made rectangles some of the time. 4. Computation of Gaussian - it took exp (+ R ** 4) rather than exp (- R**2)! 5. Changed it to use CCINI so that the CC file would have column labels, units etc to allow PRTCC to access it. 6. Fixed handling of BITER - it could fail because one tried to write well after pre-existing portions. 7. Changed help file - units are pixels, BITER causes over-writing of pre-existing components. Moved nowhere. 9836. April 16, 1998 COMB Eric A bug kept the output Stokes from being set to something besides the input when appropriate. Moved nowhere. 9837. April 16, 1998 TKSET adverb Chris Added calls to ZTKOPN and ZTKCLS around the call to SLOCIN in AU9C. This prevents TKSET and TK1SET failing because the Tektronix LUN is not open. Moved nowhere. 9838. April 16, 1998 CHTRIM in APLSUB Chris Added code to prevent CHTRIM writing beyond the end of the output string if the trimmed input string is longer than the output variable. This was causing failures in FITLD. The string will now be truncated if it doesn't fit the output variable. Moved nowhere. 9839. April 17, 1998 UVDATA object Chris OUVSCR was issuing a size request to SCREAT in units of AIPS bytes rather than block. This was causing BLING (and perhaps other tasks) to create absurdly large scratch files. Moved nowhere. 9840. April 17, 1998 BLING Chris Fixed a number of errors. 1 - Two FORMAT statements in WARNST contained concatenation operators instead of commas. 2 - The KEYIN routine that reads the auxilliary file expected ALL as a token instead of ANY. 3 - Acceleration interpolation multiplied the adjustment in cells by the acceleration spacing twice resulting in a negligable change from the cell centre. Moved nowhere. 9841. April 19, 1998 FITLD Eric **** FITLD was making source tables with multiple occurrences of essentially the same source and with multiple occurrences of the same source number attached to different sources. ***** The VLBA Correlator treats the same source, with different CALCode as a different source. This causes AIPS to have things in the source table that it cannot properly address and is simply due to the VLBA antennas passing on CALCodes while non-VLBA ones do not. Changed FITLD to allow for source renumbering even while reading the first input file and changed the list handling to include current additions to the source list when checking the next input source. The logic was backwards in handling the source number translation, complicating the process and forbidding any case in which 2 or more source numbers would both be translated into the same output source number. This arises: file 1 had a cal as source 1, 3C120 as sources 2 and 3 (w different CALCode). File 2 has 3C120 as sources 1 and 3 with the cal as source 2, so both 1 and 3 have to become 2 on output. Also fixed up the formatting of times. Moved to 15APR98 patch area this date. 9842. April 20, 1998 FITLD Chris Added code to FLTAB in APLNOT and to FITLD itself so that VLBA flag tables with conforming to table revision 1 (as documented in VLBA Memo 108) can be read in addition to the undocumented table revision 1. Flag tables from Penticton should now be read correctly. This is a band-aid type fix: routines that with VLBA tables don't seem to be well suited to dealing with multiple revision levels (which will be a necessity with 3 correlators providing data in VLBA format). Moved nowhere. 9843. April 21, 1998 GAL Eric For some reason GAL did not clean up after itself when it failed, did not test for zero divides in a model computation routine, assumed that the deprojected velocities are always positive, and otherwise got itself into trouble. Fixed it up a little. Moved nowhere. 9844. April 21, 1998 FITLD Eric For some reason FITLD chose to stop reporting new scans after the USUBA condition was raised under most but not all conditions. Changed it to report new scans all the time. Also added tests and warning/error messages if troubles arise in the source table for same source with two numbers and same number with two sources. Moved nowhere. 9845. April 22, 1998 IMAGR Eric In the 3D imaging mode, the routine doing the major cycle, one field at a time, has used only the count of iterations in that field to set cutoffs and the like. I changed CLACLN to compute the total number of iterations over all fields and to maintain and use that total while Cleaning each field. Moved nowhere. 9846. April 23, 1998 ZTKSRV Eric ZTKSRV has $APLDEC and $APLSGI versions which were overlooked when I added the image catalog code to the $APLUNIX version. Cleaned this up. Moved to the 15APR98 patch area this date. 9847. April 23, 1998 XAS Eric The cursor was coming out black (on black) on hominid, our latest DEC machine. It turns out that the color code for this machine (only) requires the colors to be in the low bits, all others require them to be left shifted 8 bits (including our older DEC machines and PCs - this is not a byte order problem). Changed XAS.SHR to put the cursor color in both the low and high bytes. Moved nowhere. 9848. April 23, 1998 $APGNOT/PCCOR.FOR Leonia PCCOR has read the first row of the PC table to take the frequencies of the tones. It is wrong for multiple freqid case. I modified it to read the first row of the first appearence of the given freqid. The maximum and minimum tones for each IFs are calculated now instead of taking the first one and the last one. The dimensions of the arrays NANPC and DCABL are changed to MXAN from 15. The value of the parameter MXAN is increased from 20 to 30. Moved nowhere. 9849. April 23, 1998 FITLD Eric Reinstated my short-lived DFILL argument fix in GETDEL. Changed KBUFFA from 256 to 512 words in FITLD since it is used for a table buffer. These caused aborts on HPs. Moved nowhere. 9850. April 24, 1998 LDGPS Chris Fixed LDGPS to handle dates in the YYYYMMDD format. Also removed some ad hoc elements from the calculation of GPS time and added some protection against the 1024 week rollover (this happens for the first time on Aug 22 1999). Moved nowhere. 9851. April 27, 1998 QEDIUTIL Eric An addressing bug caused the table editing to be unable to reach the very last sample in the data set. This addressing was handled correctly when UV data were read as in EDITR and SCMAP. Moved nowhere. 9852. April 27, 1998 FITLD Chris FITLD was reading what it presumed to be the first data record of each UV_DATA table without first checking if the table was empty. In the case that the table was empty, this lead to the first record of the header of the next table being consumed and to FITLD losing its place in the file. FITLD now checks the number of rows in the table before reading data. Moved nowhere. 9853. April 28, 1998 UVFLG Leonia Spurious fringers appear due to pulse cal injection, if fringe rate close to zero. These spurios fringes are self averaged at fringe rate stopping procedure if fringe rate is not close to zero. Time interval required to suppress the spurious fringes is calculated by Kogan at VLBA Test memo no 58, 1998. The new routine FLGPCL has been added to flag time interval at the vicinty of calculated time where fringe rate is zero. The length of this interval depends on the suppression factor. (APARM(5)). APARM(6)>0 allows to print source, time, baseline, and U,V of expected zero fringe rate (U=0) So the user can pay a special attention to the PCAL tones at the vicinity of the points. The help file is updated. Moved nowhere. 9854. April 28, 1998 ACCOR Leonia The logical error has been found inside of routine ACCRCOR. This error brought in some extra points at the data gaps if SOLINT less than the gap's length. Moved nowhere. 9855. April 28, 1998 DBCON Eric The code for rewighting the data had an off by one error in addressing the weights when the two data sets are in the same order. So the rewighting was done to the imaginary parts of the vis! Moved to the patch area this date. 9856. April 29, 1998 BP cal Eric If an antenna is missing from a BP file entirely, but present in the data, then it is possible to have BPGET return with nothing set. Changed BPGET to return IERR=-1 in this case. Changed DATBND to handle this by flagging all of the data. Upgraded the counting of partial and full flags as well. IOBSRC did not set its flags if the data were not found - so they were left at their initial values. Moved nowhere. 9857. April 29, 1998 $AIPNOT/SETPAR.FOR Jeff Removed restriction that CRTMAX must be between 9 and 200; negative values will soon be used to activate auto-detection of screen size. Moved nowhere. 9858. April 29, 1998 $INC/AIPSPROTO.H Jeff (Also a change on 1998/04/15.) Added prototypes for SIGWINCH handling. Moved nowhere. 9859. April 30, 1998 $APGOOP/COHER.FOR Leonia A small mistake has been found in baseline selection. It is fixed. Moved nowhere. 9860. April 30, 1998 PROC helps Eric Added comments to PROC.HLP and PROCEDUR.HLP concerning the choice of procedure name. Bad choices can confuse the parser. Thus proc FRINGIT causes GO FRING to become TPUT FRINGIT, FRINGIT. POPS does not know task names (thank GOD!) and so can be confused by bad verb and proc names. Moved nowhere. 9861. April 30, 1998 FLGIT Eric New task to apply flags based on a UVMLN like algorithm to corrupt (bad RFI) spectral-line data. It needs a decent BP to fit a linear Real and Imaginary to look for deviant points, but actually does seem to do some good. Moved nowhere. 9862. May 1, 1998 TVFLG, SPFLG Eric The number of TV color levels was not initialized before the call to TVPSUD. This caused the loops in hue to behave badly. Moved nowhere. 9863. May 1, 1998 UVMLN Eric In doing FIXWT, I found a funny startup message and a bunch of unused variables to remove from UVMLN. Moved nowhere. 9864. May 1, 1998 PCNTR Eric Corrected a positioning error for text lines below the plot. When there was no rotation, the line was placed one too high. Moved nowhere. 9865. May 4, 1998 COMRPL, COMLNK Pat Put in a few vital lines that make it possible for users to compile and link tasks in their private areas without needing write access to the PREP area. If the $PREP directory is not writable by the user, PREP is temporarily redefined to point at the current working directory. Moved to the AOC this date (so Bryan Butler can use it). 9866. May 4, 1998 VPLOT.HLP Leonia Regime of editing bad amplitude points (APARM(1)) works naturally only with plotting of amplitude. This means that BPARM(2) should be equal 1 or its default 0. This statement in explicit form is added. Moved nowhere. 9867. May 5, 1998 BPASS Eric Found several serious errors: 1. The worst was an addressing error which meant that the address for weights did not get incremented when a channel was flagged. Then the weights for higher numbered channels were put in for lower numbered ones. This could lead to channels being eliminated, but did not seem to cause flagged channels to acquire false solutions. The excess channel elimination did lead to a message, so users should have noticed. 2. The fit routine did not return an error code for no valid samples so the task did not know to invoke interpolation when a channel was fully flagged. 3. The interpolation method used the wrong formula for weighting channels so that the interpolation was not quite right and it would fail to extrapolate a solution if the last couple of channels or more were failed. Moved to the patch area this date. 9868. May 5, 1998 UVUTIL eric The UVREFQ routine attempted to convert the reference pixel and frequency to account for BCHAN and channel averaging but did not do it quite correctly. The only routine to call it U2IDES assumed - in the line case but not the continuum case - that no correction had been made and then attempted to do it right. Unfortunately, it was doing it to a partially corrected value. No error occurred if BCHAN and the input uv data reference pixel were both 1 which is a common case or if no (or all) channels were being averaged. Moved to the patch area, sigh.... 9869. May 6, 1998 Autocorrelation phases Eric Yes baseline n-n may have phase, but only in RL or LR. So changed: FITLD to handle the RL and LR in the correction routines as cross correlations rather than autocorrelations PRTUV to allow a request for baeline n-n in CPARM(9), to do amp and phase of n-n RL and LR as cross correlation. POSSM to allow it to plot phase if STOKES "RL' or 'LR' in auto mode, dropped APARM(8)=5 setting that for RL and LR, fixed error in selection which left phases 0. POSSM HLP - dropped APARM(8). Moved nowhere. 9870. May 6, 1998 Various $SYSUNIX scripts Pat The initial work to enable the *optional* use of UNIX sockets for communications between AIPS tasks and the various graphical servers (XAS, TEKSRV, MSGSRV, TVSERV) has now been fully put back in the system. This functionality is parallel to the existing INET socket scheme that AIPS has been using since 1992. While not in its final state, the use of UNIX sockets in this manner is now at a point where it can be tested in various ways. Multiple instances of the XAS TV and the other servers are now possible on a single host; this has been tested and verified on Linux/Intel and Solaris/Sparc platforms. The feature is *only* enabled via the USOCK and optional UNEW options to the AIPS command line. These names are not cast in stone, and their usage may also change. Known problems: 1. After starting up one or more Unix socket based TV's on a given host, it is impossible to fire up an INET socket based TV. The workaround is to start up the INET socket TV first. Solution is to modify search patterns in TVDEVS.SH to anchor the 'grep' for XAS at end of line. 2. With multiple instances of the TV on the same host and display, it is currently impossible to select which one you want to use for new AIPS sessions. A solution would be to modify the syntax of the command line (e.g. USOCK=n) to allow the user to choose, and modify the scripts accordingly. 3. Detection of pathological cases has not been implemented. Example: XAS for USOCK session 1 died but the other servers (MSGSRV, TEKSRV, TVSERV) did not. The startup scripts will not notice the latter servers and will always use slot 1 (assuming the display host matches) even if UNEW has been specified. Files affected: START_AIPS: implement USOCK, UNEW. This file needs copied to $AIPS_ROOT (done in CV) and its AIPS_ROOT definition changed as appropriate to enable the feature. The AOC will hold off on doing this pending further tests. START_TVSERVERS: ditto. Calls UNIXSERVERS or XASERVERS. TVDEVS.SH: New logic to figure out TV number. MNJ sites should check if $AIPS_ROOT/TVDEVS.SH is a symlink to the wrong version of $SYSUNIX/TVDEVS.SH and fix accordingly. UNIXSERVERS: needed cleaned up somewhat. Moved nowhere. 9871. May 8 $APGNOT/PCCOR.FOR Leonia There was a mistake in phase referencing. It was for the first tone frequency, not for the referencing frequency. I fixed it to reference to the reference frequency. Moved nowhere. 9872. May 8, 1998 FITLD Eric Changed the help file to drop CALCODE - I changed the code for this two days ago. It worked in my area, but not when released. CALCODE is unpredictable and should not be used for selections anywhere near FITLD. Moved to the AOC by hand this date. 9873. May 17, 1998 0-length baselines Eric Changed more things to allow phases on RL, LR, XY, and YX correlators: UVPRT Changed to have a logical control whether amplitude and phase of real part (signed) value is displayed. PRTUV Changed to test for RL and LR more carefully than just correlator 3 and 4. UVPLT Changed to handle autocorrelations without forcing the amplitude to be positive and to recognize RL and LR as having phase even when IA1 = IA2. Moved nowhere. 9874. May 17, 1998 REFRMT Eric Wrote a new subroutine to compress consecutive blanks out of a display string. Now to display channel number use a generous format like I5 and then call REFRMT. If you want more than one consecutive blank in some places in the string, place a "special character" (actually some real one like underscore) where you would like a blank that is not counted as a blank. Also changed the handling of the LEVS list in COMLAB, GREYS, and KNTR. Changed to use REFRMT and to handle the COMLAB change COMLAB PLAXES PFPL1 PFPL2 PFPL3 PLROW PLCUB IMRMS UVPLT ISPEC IMVIM SL2PL TAPLT UVHGM BPERR BSTST IMEAN SERCH BPLOT CLPLT FRPLT SNPLT FRMAP ELINT VPLOT POSSM DFTPL GAL IRING TVFLG SPFLG IAXIS1 KNTR GREYS CNTR CCNTR PCNTR Also changed the display in ACOUNT to give one more character for cpu time when IOCNT is kept and several more, leaving out null IOCNTs when not. Moved nowhere. 9875. May 17, 1998 ANTENNAS, BASELINE Eric Changed: SETANT $APLSUB - new to convert ANTENNAS and BASELINE into counts of unique and DESEL on either. REQBAS $APLSUB - to simplify how DESEL is coded and to have BASELINE and ANTENNAS be symmetric. AN10RS $APLNOT - to use SETANT and to deselect MAXANT-MAXANT as well as the other autocorrelations which were done before. BASLST $APLNOT - to use SETANT. ANTENNAS $HLPFIL - to clarify the meaning and to add info re BASELINE BASELINE $HLPFIL - to correct the wrong statements and add the same info about the interaction with ANTENNAS. FRMAP $YPGNOT and $HLPFIL - to remove special REQBAS and rearrange the handling of ANTEN and BASEL, using SETANT VPLOT $YPGNOT and $HLPFIL - to remove special REQBAS and rearrange the handling of ANTEN and BASEL, using SETANT COHER $YPGNOT and $HLPFIL - to remove special REQBAS and rearrange the handling of ANTEN and BASEL, using SETANT UVPLT $YPGNOT - to have BASEL set deselect too. POSSM $YPGNOT and $HLPFIL - clean up antennas/baseline BLING $QPGOOP - to deselect on BASELINE too. UVCOP $APGNOT - history routines did not handle deselection and were messy - changed to use SETANT. Also reworded the help file some. UVIMG $APGNOT - changed to use REQBAS rather than home brew and to use SETANT. GRIDTB $APLNOT - changed to use REQBAS rather than home brew To use SETANT: UVPRM, UVPRT, CLIP, IBLED, TVFLG, SPFLG Moved nowhere. 9876. May 20, 1998 XASERVERS, UNIXSERVERS Pat The methods by which the UNIX socket based TV's are started up has changed slightly: AIPS TV=LOCAL forces Use of UNIX sockets AIPS TV=LOCAL:0 always creates a new instance of the TV AIPS TV=LOCAL:n use local tv #n (n=1,2,...), or start it if it's not already running This does not affect the existing INET based TV functionality at all; it will continue to work as it did (or did not!) before this change was made. Some minor changes in the scripts were needed in order to avoid confusion in the "ps" information mining between the two types of TV. Also, the string in the title bar and the icon label now indicates what type of TV one has, and in addition which number it is in the case of a UNIX socket based ("local") TV. Moved nowhere. 9877. May 20, 1998 FLGIT Eric Added more clip parameters to go after real and imaginary as well as amplitude. Moved nowhere. 9878. May 21, 1998 TVSERV, XAS Eric Added a die command to ZTVSRV and added code to XAS.SHR to kill the TV lock server when XAS exits. The zssslk version in XAS.SHR did not prepare data for sending to the server on Unix sockets just Inet sockets. Moved nowhere. 9879. May 21, 1998 XAS Chris Inserted code to shutdown and close AipsSocket in closedown(). This prevents the lingering "socket in use" problem under Solaris. Moved nowhere. 9880. May 22, 1998 FLGIT Eric Added a new mode to subtract a median window value from each real and imaginary and the clip and flag from there. This may reduce the need for bandpass calibration. Moved nowhere. 9881. May 22, 1998 TVSERV Chris Moved the socket clean-up code out of the main loop in ZTVSRV and forced an exit from the loop when requested to shutdown by XAS. Closing down the sockets and exiting from inside the loop appears to cause lingering "socket in use" problems under Solaris; it is not clear why this should be the case. Moved nowhere. 9882. May 24 & 26, 1998 Servers Eric Took Chris's suggestion and added finished to ZTVSRV, ZMSSRV, ZTKSRV so that loops will be exited and close down calls will be made when errors occur (or when TVSERV is ordered to close). Dumb error in ZTVSRV (by me) fixed on 26 May. Moved nowhere. 9883. May 26, 1998 SNPLT Eric Added RATFAC scaling to the one rate plot that did not already have it at suggestion of Michael Bietenholz. Moved nowhere. 9884. May 26, 1998 FRING Eric At the suggestion of Michael Bietenholz I have changed FRING to call FRNADJ when DPARM(7) <= 0 or DPARM(8) > 0. Inside, that routine I added DOREF=DPARM(7).LE.0 and test on it before doing the re-referncing parts. Moved nowhere. 9885. May 27, 1998 IMFLT Eric Fixed bug I introduced in 1995 (!) which made the output no good - the input buffer was 4096 while MINIT was told it was 8192. Corrected handling of BOX defaults, introduced PIXRANGE to limit the pixels used in the fit, cleaned up the history handling, etc Moved nowhere. 9886. May 27, 1998 PCCOR Leonia The standard AIPS parameters MAXANT and MAXIF have been introduced instead of local MXAN and MXSIF. Moved nowhere. 9887. May 28, 1998 VLOG Eric Added tests for bad characters and Nan's substituting ? and 999 and issuing a message. This is done on all input card images. Moved nowhere. 9888. May 28, 1998 $APLDEC/ZFRE2 Pat Modified this to use the actual number of bytes per block as returned by the OSF1 statfs() call; it was using a fixed value of 1024 bytes. This resulted in Astronomers on Alpha systems thinking they had twice as much space as they really did. The old assumption may have been necessary on OSF/1 1.2 but not on any recent incarnation of the OS. Moved nowhere. 9889. May 28, 1998 STARS Eric The position angle of 0 degrees has the major axis along the y axis, STARS acted as if the major axis goes with the x axis. Switched these and added tests to force PA=0 and to avoid precessing velocities as if they were angles when the two first axes are not a position pair. Moved nowhere. 9890. May 28, 1998 TVSERV Eric Changed ZTVSRV to suppress SIGINT. The CTRL-Y to AIPS was passing through and killing TVSERV (but not XAS, MSGSRV, TEKSRV). Moved nowhere. 9891. May 29, 1998 CCNTR Eric Add a SETLOC call ahead of the CC and ST plotting to restore the plot parameters to standard rather than label-dependent. Moved nowhere. 9892. June 1, 1998 $APLDEC/ZTAP2.C Eric Added tests for being at BOT at the start of the operation and use this to avoid BAXF, BAKR, REWI etc from there. Added tests for wierd set of error conditions from a BAKF that hits the BOT and calls to reset to error condition status (later reads still did not work even though the error condition appeared to be ok). Moved to siamang for testing. 9893. June 2, 1998 Y2K Eric Checking things at Tony's insistence: SETAN Increased string lengths for KEYIN, call DATFST to translate either form to internal ("hand" translation of old form was there). GETKEY Gets table keywords but needs to test for long string of form DATE.... plus RDATE and UTDATE and call DATFST before returning the 8-char string FITTP Put header and table keywords in new FITS date format. Moved nowhere. 9894. June 3, 1998 TABHDR Chris Changed to 2-letter table codes for gain curve, phase-cal and system temperature tables that have been read from FITS interferometry data interchange files to GN, PH, and TS respectively. This prevents them being confused with the equivalent tables used internally by the AIPS calibration system. This has the side effect of temporarily disabling calibration transfer until changes to FITLD are complete. Moved nowhere. 9895. June 3, 1998 ZTAP2.C Eric Added another magic number to the BAKF at BOT tests - the write-lock bit affected the magic value previously tested. Moved to Bologna this date. 9896. June 3, 1998 TABHDR Eric Add DATFST call on date string to Chris's changed version. Moved nowhere. 9897. June 3, 1998 OTFUV Eric Changed directory handling so that the OFF and GAIN scan numbers are found in the data and then sought in the directories. In this way, BCOUNT amd ECOUNT do not have to apply to them. Moved nowhere. 9898. June 4, 1998 START_AIPS Pat Of all the calls to uname -n, one did not have the secondary awk code that converted a fully qualified domain name [FQDN] (e.g. orangutan.cv.nrao.edu) to a simple hostname (orangutan). On systems where uname -n returned a FQDN, this caused the local unix-socket tv startup to fail. Midnight job sites should move this to $AIPS_ROOT. Moved to AIPS_ROOT in Charlottesville today. 9899. June 5, 1998 GETKEY Eric Changed it to call DATFST for any character-valued keyword with DATE in the symbol. Moved to the patch area this date. 9900. June 9, 1998 CPASS.FOR Leonia CPASS has produced additional column (weight) at the output BP table starting with the version 15APR. The weight calculated correctely only for the first polarization (LL) and the first IF. The weight (calculated in the routine BPUPD) has been zero for all other polarizations and IFs. As a result the new version of POSSM which uses the weights ploted only the first polarization and the first IF. The problem has been fixed. Moved nowhere. 9901. June 10, 1998 OTFUV Eric An off by one failure caused muti-IF runs to fail. Moved nowhere. 9902. June 10, 1998 OTBSRT Bill/Eric Did not return an error code when it failed! Calling routines then assumed the output table == input table was sorted as needed when it was not. Results are more than a little unpredictable. Moved nowhere. 9903. June 11, 1998 LPCAL Eric An 8-character HOLLERITH was declared as a single Hollerith. This led to an abort on HP computers. Moved to MPI-Bonn and to U of Bordeaux this date. 9904. June 11, 1998 SPLIT Eric SPLIT took NOPCAL to be no of parameters per IF and so cleared an array of size NOPCAL * MAXIF. Unfortunately, NOPCAL is actually 2 * Number of IFs already. Moved to MPI-Bonn this date. 9905. June 18, 1998 FITLD Chris Overhauled the handling of calibration tables from FITS Interferometry Data Interchange files: specifically the FLAG, GAIN_CURVE, PHASE-CAL and SYSTEM_TEMPERATURE tables. These interchange files are now regarded as distinct from their AIPS equivalents and are checked to make sure that 1) keywords that are necessary for the interpretation of the table are all present, 2) the required columns are present, 3) the required columns have the correct types, and 4) the required columns have the correct dimensions. If any of these checks fails then a diagnostic message will be generated and the table will not be translated to its AIPS equivalent but will be retained in its original form. In their original forms, the table have two-letter codes that are distinct from the eqivalent AIPS tables so that faulty tables will not interfere with later processing. The assigned codes are FLAG FL (AIPS equivalent FG) GAIN_CURVE GN (AIPS equivalent GC) PHASE-CAL PH (AIPS equivalent PH) SYSTEM_TEMPERATURE TS (AIPS equivalent TY) These changes are intended to decouple the AIPS tables from the interchange tables so that they can be revised separately and to allow FITLD more flexibility in dealing with multiple table revisions in interchange files as well as improving the handling of faulty tables. The following interfaces have been added to deal with the interchange tables. INC:PFLV.INC (replaces DFLV.INC) INC:PGNV.INC INC:PPHV.INC INC:PTSV.INC APLNOT:FLINI.FOR APLNOT:GNINI.FOR APLNOT:PHINI.FOR APLNOT:TABFL.FOR (replaces FLTAB.FOR) APLNOT:TABGN.FOR APLNOT:TABPH.FOR APLNOT:TABTS.FOR APLNOT:TSINI.FOR The following interfaces have been modified to remove keywords that are only needed for data interchange from internal AIPS tables (where they are not needed) and to make table access routines conform to normal AIPS table access conventions. INC:PPCV.INC (replaces DPCV.INC) APLNOT:GCINI.FOR APLNOT:PCINI.FOR APLNOT:TABPC.FOR (replaces PCTAB.FOR) The following routines and tasks have been modified to use the revised GC and PC table routines APLNOT:GCSEL.FOR APLNOT:PCSEL.FOR APGNOT:ANTAB.FOR APGNOT:PCCOR.FOR APGNOT:PCLOD.FOR YPGNOT:APCAL.FOR The following files are now obsolete and have been removed from the system INC:DFLV.INC INC:DPCV.INC APLNOT:FLIO.FOR APLNOT:FLTAB.FOR APLNOT:GCKEY.FOR APLNOT:PCIO.FOR APLNOT:PCTAB.FOR NOTE: the removal of keywords from the GC and PC tables means that GC and PC tables from 15OCT98 AIPS can not be used in earlier versions of AIPS. Moved nowhere. 9906. June 18, 1998 BPASS Eric My formula for interpolation was backwards. Moved to the patch area this date, patching the patch. 9907. June 22, 1998 PCCOR Chris Fixed a number of errors that I had introduced while updating the PC table interfaces (mostly 'WRIT' opcodes where there should have been 'READ' opcodes). Moved nowhere. 9908. June 22, 1998 FIXWT Chris Changed MEDIAN to use a temporary copy of the data buffer so that the data buffer is unmodified in the call as expected by the higher-level routines. Moved nowhere. 9909. June 22, 1998 TABPC Chris The indexing calculations used to write data to the data buffer incorrectly used the constant 4 where NUMTON (the number of tones) should have been used. Moved nowhere. 9910. June 24, 1998 TASKWT Eric Changed AIPS to wait at least 15 seconds for a task to start. Fast machines are having problems when served by slower machines or networks. Moved nowhere. 9911. June 28, 1998 OTFUV Eric There was an error in the new code addressing the gain file when the scan numbers for the offs were very different from the scan numbers for the gains. Moved nowhere. 9912. June 29, 1998 FLGIT Eric Corrected it to write SU tables only if the output file is multi-source. Great confusion (see below) reigns otherwise. Moved nowhere. 9913. July 6, 1998 FITLD Chris FITLD subtracted the total bandwidth of an IF from the reference frequency of a lower sideband IF when swapping the channels into ascending order of frequency. This is only correct in the special case where the reference pixel is 0.5. The subroutines that deal with the reassignment of LSB frequencies, FRQORD and FR2FQ, have been modified to deal with the general case. In addition, reference frequencies are no longer rounded to even multiples of 10 kHz, which lead to shifts of a few kHz in the labelling of LSB frequencies. These changes copied to 15APR98 patch area on this date. 9914. July 7, 1998 SHOW & TELL Eric Changed the diensionality of the SHOW & TELL file (TC). It was handled as if the maximum POPS number was 15 although is is now 35 and it probably did not have enough entries as well. Better clean up of pending TELLs is needed also. So: AU2 Corrected error that caused it to use the next POPS number when one was full. Changed addressing by factor 2. Changed STQ to clean up inactive TELLs always - I assume that ZTACTQ now works pretty reliably. Changed TC file name to TCD000001 GTTELL Changed addressing. Changed to use NPARM < 0 to mean "clean out all entries for this task/POPS" for any user number. New TC file name. GTPARM Add call to GTTELL (-1...) DIETSK Add call to GTTELL (-1...) FILINI Changed the INIT section for new sizes of TC New TC file name FILAIP Changed file size to 18 + 16*(NINTRN+2+NBATQS) New TC file name FILAI2 Changed file size to 18 + 16*(NINTRN+2+NBATQS) New TC file name STQUEUE Help file updated to drop special clean up option. Moved nowhere. 9915. July 7, 1998 IMHEAD Eric Change LSTHDR and MSGHDR to use 2 more digits each for shifts - the 74 MHz images require it. Moved nowhere. 9916. July 7, 1998 DATGET Eric Put it back to the old convention of not returning fully flagged data. For a while, if DOFLAG was false, it would return fully flagged data. Upper level programs did not expect this and made errors. Moved nowhere. 9917. July 7, 1998 Multi-source detection Eric Changed: MULSDB To check whether ILOCSU >= 0 as well as SU table exists. Corrected to use ILOCSU >= 1 in OOP: COHER, MAPBM, UVUTIL, UVPHAS. Also added tests for using FQ and SU to COHER and MAPBM AND UVUTIL. Corrected to use ILOCSO and ILOCFQ >= 0 for regular code: CALIB PCAL LPCAL CPASS FRING VPLOT CLPLT UVAVG PHSRF FXVLA SPCAL LISTR UVFND BLAVG Corrected logic to set MULTI includING ILOCSU and force ILOCSU >= 0: BLCAL, BPASS, CALIB, LPCAL, PCAL, FRING, SPCAL, CVEL Added tests for valid ILOCSU, ILOCFQ: ACCOR, PCCOR, UVHGM Moved nowhere. 9918. July 7, 1998 DFLG.INC Eric Pulled the flagging stuff from DSEL.INC, creating separate DFLG.INC. DOFLAG and FGVER are left in DSEL. Also pulled the parameter MAXFLG from PUVD to PFLG.INC. Changed: AVSPC Changed local flag include to be mostly DFLG.INC. UVCOP Made a local include for PFLG with 50000 simultaneous flags allowed and appended FLGINI, DATFLG, and NXTFLG modified to use the local include. UVFLG Put PUVD.INC in the task include, simplifying things. UVMLN Added PFLG/DFLG and fixed a declaration DATFLG Added PFLG/DFLG. FGREFM Fix comment. Simply added PFLG and DFLG to: NXTFLG SDGET UVGET CVEL SDVEL IBLED SPFLG TVFLG VPLOT Moved nowhere. 9919. July 7, 1998 DHDR.INC Eric As long as I am breaking everything, add a header parameter for a normalization factor for uv data weights. DHDR.INC Added KRWTN VHDRIN Added KRWTN at the end of ext files CATN2L Added KRWTN translation back and forth CATINI CATR(KRWTN) init as 1.0 LSTHDR Changed to report non-0 non-1 normalization factors. MSGHDR Changed to report non-0 non-1 normalization factors. Moved nowhere. 9920. July 7, 1998 Also use DFLG Eric Several things escaped my GREPs since they use KNCS etc and hence need DFLG/PFLG as well as DSEL: BASAVG DATBND DATCAL DATPOL DCALSD SMOSP CLPLT In the case of SOUFIL, found a buffer which was too large equivalenced to FGBUFF which could cause all sorts of problems for the unwary! LISTR used FGBUFF for scratch - changed to BLBUFF. Moved nowhere. 9921. July 7, 1998 SDVEL Eric Corrected the DFLG business and added/changed options having to do with when and with what coordinate the correction is computed. Also the help file. Moved nowhere. 9922. July 7, 1998 FLATN Eric Corrected bugs that (1) caused the center position to be that of field 1 rather than the pointing position, and (2) caused it to check for the existence of fields 2-N only when actually opening them for interpolation. The latter exposed a foible of OOP - one can ask for header parameters of non-existent files (or existing ones I suppose) that have not been opened with OOPEN and get back zeros for both the parameter and the error code. Added COORDINA adverb to the task to allow the user to set the central coordinate and fixed it to zero the shift parameters and put the sum of the numbers of CC components into the header. Dropped excess calls to copt observed coordinates and beam parameters since that is all done in IMDCOP. Moved nowhere. 9923. July 9, 1998 ZABOR2 in APLUNIX Chris Removed the definition of the _POSIX_SOURCE feature test macro from ZABOR2 in APLUNIX. The introduction of SIGWINCH into this module means that it no longer complies with POSIX standards (or any other UNIX standards, for that matter) and the SGI compiler correctly refuses to compile it when the _POSIX_SOURCE macro is defined. Moved nowhere. 9924. July 9, 1998 $APGNOT/PCCOR.FOR Leonia The frequencies of the tones have been determined from the first row of the PC table (for the given selected parameters). PCCOR failed if this row is bad (filled by zeros). I added the codes making PCCOR look through the first succesful row. Moved nowhere. 9925. July 9, 1998 GAININ Eric GAININ checked for empty CL/SN tables and set DOCAL false. Unfortunately, it did not say anything and did not close the file thereby leaving a much used LUN occupied. Moved nowhere. 9926. July 10, 1998 ZWINCH in APLUNIX Chris Removed a spurious definition of _POSIX_SOURCE (see the ZABOR2 change above). Moved nowhere. 9927. July 11, 1998 $APGNOT/PCCOR.FOR Leonia PCCOR uses the two tones (the biggest and the lowest frequency) even if more than two tones present at each IF. Originally I supposed that the last tone has the biggest frequency. It seems not true. Currentely the second tone has the biggest frequency. As such a sequence of the tones is not fixed forever, I modifyed PCCOR recentely to calculate the the lowest and biggest frequency tones. Unfortunately I broke unwittingly the logic for USB/LSB IFs during those modification. I fixed it now. Moved nowhere. 9928. July 14, 1998 SIGWINCH handling Jeff Finished adding SIGWINCH detection for LINUX, DEC, & SOL a few days ago, also added "stub" for other OS's. Files added: $APLGEN/ZWINC.FOR: called by ZDCHIN & ZWINCH to call ZWINC2 and get the current window size. Sets CRTMAX based on ZWINC2's return value. $APLUNIX/ZWINCH.C: signal handler for SIGWINCH, calls ZWINC. $APLGEN/ZWINC2.FOR: stub for OS's that haven't had SIGWINCH support explicitly added for AIPS; returns 0 which tells ZWINC to leave CRTMAX's value alone. $APLLINUX/ZWINC2.C, $APLSOL/ZWINC2.C, $APLDEC/ZWINC2.C: OS-specific routines for determining window size. Called by ZWINC. Files modified: $APLGEN/ZDCHIN.FOR: Added call to ZWINC to get initial window size & set CRTMAX. $APLUNIX/ZABOR2.C, $APLSOL/ZABOR2.C, $APLDEC/ZABOR2.C: ZWINCH as handler for SIGWINCH. Chris has already fixed up a couple of things I'd not thought of: _POSIX_SOURCE tests/sets in $APLUNIX/ZABOR2.C and $APLUNIX/ZWINCH.C. Moved nowhere. 9929. July 15, 1998 FITTP Eric Dropped the DOTWO adverb since it no longer can do anything. Also added remarks in the help file to point out that uv data is only IEEE floating point. Changed initial values in POPSDAT.HLP for DOTWO to +1, for DOCRT to +132, for DOCALIB to -1. Dropped FITTP, added EDITA and EDITR to DOTWO.HLP, Moved nowhere. 9930. July 16, 1998 CALIB.FOR Leonia CALIB has not worked with multiple source file if model is CC file of CLEAN map. I found and fixed two bugs both in subroutine SETFAC, which calculate the ratio of the flux from CC table and flux from SU table. The first bug: The output parameter of subroutine CATKEY - KEYTYP was fourced to be 2. It lead to the confusion with memory. The second bug: Reading SU table the row number was equal the source ID number. So the task can not work if the first row does not correspond to the source ID = 1, especially if the source ID of the first row is more than number of rows in SU table. Moved nowhere. 9931. July 16, 1998 BPASS, DATBND Eric 1. BPASS was writing the actual bandpass values when they were determined with autocorrelation only. It was then taking a sqrt in DATBND when applying to autocorrelations. This is fine when the BP was found and applies only to autocorrelations. But is not if it is found with cross correlations or applies to cross correlations. Changed it to take the sqrt in BPASS instead. 2. BPASS was finding the max antenna number by an ANTINI call and the AN file record count! Wrote ANMAXA subroutine to read the whole file and look at the an numbers in each record to see what the max really is. Many more tasks are involved - not tonight! Moved nowhere. 9932. July 19, 1998 CRT printing Eric Changed PRTLIN and PRTALN to call ZWINC after the page full do we continue query. Changed LPOPEN to call ZWINC after opening the TT device for DOCRT > 0). Added calls to ZWINC also to AU1A, AU7, OFMLST, and PRTMSG since in $AIPSUB LPOPEN is not used for DOCRT > 0. Relink every task compiled on and since July 8 since they included ZABOR2 changes Moved nowhere. 9933. July 20, 1998 FXPOL Chris Added logic to convert gain curve (GC), phase-cal (PC) and system temperature (TY) table to dual polarization tables. This is necessary for tables loaded from VLBA distribution tapes if RR and LL are present without the cross polarizations. Moved nowhere. 9934. July 20, 1998 $YPGNOT/CONFI.FOR .HLP Leonia The task finds the array configuration which has the minimum worst side lobe at the given circular area at the sky. Initial configuration is taken from the input file or calculated as a homogeneous configuration on the circumference; a configuration on the circumference with progressively increasing of the spacing; a homogeneous configuration on the two circumferences; or a homogeneous configuration on the three circumference; The optimization is provided under the following constraints: constraint of topography constraint of donuts constraint of two circles. The process of finding the configuration is iterational. At each iteration the position and value of the worst side lobe in the given sky area is found. Then a small correction of each element position is provided to minimize the value of the worst side lobe. The value and the direction of the correction is calculated in accordance of the theory (L. Kogan, MMA memo # 171 ). Both the found configuration and relevant UV coverage are plotted at the same plot (on TV or in a file). The coordinates of the found configuration are recorded in the output file in normalized form or in meters. The task can be usefull in design of MMA, LSA and other arrays. Moved nowhere. 9935. July 20, 1998 EDITR et al Eric Rearranged the logic of QEDIUTIL and thereby its demonstration tasks EDITR, EDITA, and SNEDT. Added DOTWO option to all three, upgraded their help files, and their run-time helps HLPEDICL (CL/SN for uv editing), HLPEDISN (editing the CL/SN itself), HLPEDITY, and HLPEDIUV. SCMAP still needs some attention perhaps - maybe just a relink will do. Added options include flag point, flag fast, next antenna, next baseline, use expert mode, and displaying a second observable from the primary antenna/baseline. Some bugs got squashed and the colors are better I think. Moved nowhere. 9936. July 20, 1998 SNPLT Eric I find it confusing that SNPLT prefers to average and plot flagged samples, even plotting blanked as zero for CL tables! Worked it over a bit to use X's rather than +'s for WT <= 0 samples and report them too. Moved nowhere. 9937. July 21, 1998 UPDCONFIG, UPDCONTROL Pat Added some functionality that should enable a "mini" midnight job to function. Such a job will only apply COMRPL and COMLNK transaction entries if they are relevant to a list of tasks ("AIPNOT:MSGSRV AIPNOT:TEKSRV AIPNOT:TVSERV" is the default; this can be overridden in the relevant section of UPDCONFIG by setting the UpdMiniPrg variable to something different. Relevance is determined strictly by LIBR.DAT. Also removed power (was at the AOC, now in CV), added mandrill and preliminary entries for the SGI satellites in Socorro. Moved nowhere. 9938. July 21, 1998 New architecture AXLINUX Pat The EGCS 1.0.3 version of gcc and g77 has been made to produce a native port of AIPS to the Alpha/Linux environment. The only non-functional part at this time is readline; a new version is probably needed in place of the frozen snapshot we distribute in READLINE.SHR. Files modified or added: AREAS.DAT updated with new areas $APLAXLIN and $SYSAXLIN ZCPU.C new, in $APLAXLIN (different from LINUX version) ZDCHI2.FOR new, in $APLAXLIN (LINUX version has ancient rev) ZDELA2.C essentially identical to LINUX version ZTPWA2.C ditto CCOPTS.SH version for $SYSAXLIN LDOPTS.SH ditto LIBR.DAT ditto CDVER.SH ditto CDVER.CSH ditto XAS.SHR Makefile adjusted to allow AXLINUX architecture FDEFAULT.SH AXLINUX section added RUN Allow NORL (no readline) so POPSGN can function. INSTEP1 Allow recognition of new architecture REGISTER ditto FTPGET Change list of files (TC, no ICD000001;), and add AXLINUX support CDSETUP Add AXLINUX, don't make symlinks to source area if doing a full-disk install. To be moved to mandrill this date, time permitting. 9939. July 21, 1998 FITLD Chris Removed some remaining instances where frequencies were rounded to 10 kHz multiples. Introduced tolerances appropriate for single-precision frequencies (reference frequencies are single- precision quantities in VLBA export tapes) into comparisons of frequency offsets. This fixes a problem which led to rounding differences causing the creation of new frequency IDs: this had been masked by the the 10 kHz rounding. Moved to patch area on this date. 9940. July 21, 1998 FITLD Chris Changed FR2FQ and SO2SU subroutines to suppress spurious IFs when a subset of IFs has been selected. Moved nowhere. 9941. July 21, 1998 SCMAP Eric Cleaned up the handling of messages about excess baselines and number vis copied so that they come out only after they are fully known instead of with the first subarray. Fixed the explain section of SCMAP to be for the current editor. Fixed a type in QEDIUTIL re DAMP ve DAMPL. Moved nowhere. 9942. July 21, 1998 Kill SIGWINCH Eric Removed all references to SIGWINCH in $APLSOL, $APLUNIX, $APLDEC versions of ZABOR2.C and removed $APLUNIX/ZWINCH.C. Even on linux these were causing aborts for interrupted system call. Fixed missing ZWINC calls in AU1A, remembered to compile PRTLIN and PRTALN. Moved nowhere. 9943. July 23, 1998 Antenna number Eric Fixed handling of antennas numbers - or tried to: CPASS Like BPASS, it was using # records = max antenna number. Changed to call ANMAXA. FRING Like BPASS, it was using # records = max antenna number. Changed to call ANMAXA. CLPLT Like the others, it was using the number of records as the maximum antenna number. Changed it to compute the max antenna number. TVFLG It was setting locations of antennas by record number and so sorting baselines by length on poor premises. Other aspects were right due to using GETNAN. PCAL They were putting the solution for ant I in record I LPCAL even if that record and antenna J in it. This could SPCAL mess up a good polarization cal! Moved nowhere. 9944. July 24, 1998 MWFLT Eric For some reason MWFLT was different from TAFFY in handling the delayed writing of rows and so delayed one too many. Also changed it to allow for larger median filters (up to 51x51) if anyone has a mchine fast enough for that. Moved nowhere, 9945. July 24, 1998 SETSP Eric Changed to allow # CRT lines <= 0. Moved nowhere. 9946. July 24, 1998 IMLOD, UVLOD, FITLD Eric Changed to allow 64-bit primary image or uv data to come in. Corrected buffer sizes in UVLOD. Moved nowhere. 9947. July 25, 1998 BPASS.HLP Eric the inputs part said that BPASSPARM(10) had to be > 1 to get normalization while the help part said (correctly) > 0. Moved nowhere. 9948. July 27, 1998 SPLIT Eric Changed meanings of APARM(1) (frequency averaging) and added option to doing boxcar averaging of the spectrum still producing a line data set out (with fewer channels). Wrote new $APLSUB/AVGSMO to do the boxcar smoothing. Moved nowhere. 9949. July 27, 1998 TV labels Eric The positioning of labels inside the TV picture was wrong when TXINC or TYINC > 1. Also labels should not appear when the type is blank without regard for the prefix. Changed IAXIS1, ITICS. Moved nowhere. 9950. July 27, 1998 FITLD Chris Added a test to prevent unselected frequency groups being written to the output FQ table when FREQSEL is used. Moved nowhere. 9951. July 28, 1998 KLEENEX Eric New verb to exit aips cleaning out all the servers as one goes. Changed: ZTKSRV $APLUNIX/ Add new codes to force shutdown. ZMSSRV $APLUNIX/ Add new code to force shutdown. SSS.TXT $DOCTXT/ Describe new opcode to XAS XAS.SHR $YSERV/ Add new opcode to call existing shutdown routine YDIE $YSS/ Routine to open, issue die command, close SSS-type TV YDIE $YGEN/ No op. ZSSSO2 $APLUNIX/ Add new opcode to list ZSSSX2 $APLUNIX/ Avoid error message when there is no return after a die. KILSRV $AIPSUB/ New routine to issue YDIE and TEK and MSG codes for dying. AU1 $AIPSUB/ Added code for KLEENEX and cleaned up other code a bit POPSDAT $HLPFIL/ Added KLEENEX verb NEWPARMS $RUNSYS/ Added KLEENEX verb Moved nowhere. 9952. July 28, 1998 MWFLT Chris Corrected the test on the filter window to reject windows with an x dimension larger than the maximum allowed rather than the reverse. Moved nowhere. 9953. July 29, 1998 Geometry Eric Changed POSITION (PSNCV3) to do the geometry of the 3D correction wrt the tangent point and the geometry of the radial corrections wrt to the pointing position. These can be different esp in 3D imaging in IMAGR. Changed INTERPLTE routine IMGHGE to use the new call sequence of PSNCV3. Changed FLATN and its help to add the APARM adverb of OHGEO and to pass it down to IMGHGE et al. Now FLATN - for better or worse - can also apply the radial and 3D corrections. Relink OHGEO too. Moved nowhere. 9954. August 3, 1998 TABSOU, M3TAR Eric TABSOU's data buufer was 15+10*MAXIF long - the correct number was 23+10*MAXIF causing it to blow up when the full number of IFs was invoked (for some reason) by M3TAR. That task had a very silly subroutine to translate tape 64-bit numbers to local. It was a simple copy, but the offset pointer was decremented for convenience at input. Since the offset was always coded as a constant, this had the affect of changing 2, 3, etc to ?? Fortunately, some systems put constants on read-only pages and then die for this stupidity. Furthermore, simply copying the doubles is only correct for normal byte ordering not for DEC and PCs. Replaced the bad code with a call to ZR64RL. Moved nowhere. 9955. August 3, 1998 $QPGOOP/OMFIT Ketan OMFIT now offers two sets of error bars, one assuming the model is perfect (the 'suggested' error bars) and one using a thermal noise level provided by the user. The antenna-based time- independent amplitude scaling factors [gains] are now frequency and stokes dependent. The self-calibration solutions now are also frequency and stokes dependent. OMFIT can now write out the model-fit residuals as a catalogued AIPS data file. More of the input adverbs are listed in the OUTFILE. Moved nowhere. 9956. August 3, 1998 $QPGNOT/KRING Ketan KRING now has the ability to extrapolate fringe-fit solutions in time. A pre-FFT stage solution for each antenna is sought as follows. KRING looks backwards to find the last good solution whose listed SNR exceeds some multiple of the FFT SNR threshold. If such a solution is found, its delay and rate solutions are adopted for the current solution and a new phase is computed. If the SNR of this solution exceeds the FFT SNR threshold by the above-mentioned KRING skips the FFT stage. This pre-FFT stage is carried out for each antenna. After going through the entire data set, KRING will, optionally, do a 'backwards' pass in which it looks forward instead of backward for solutions to extrapolate. Unfortunately, only scans are done in backwards order. Within each scan, solutions are still sought in forward order. So, this is not a full backwards solution extrapolation Moved nowhere. 9957. August 4, 1998 $SYSUNIX/DADEVS.SH and .PL Pat By request, reworked the order in which the three categories of disk are processed. Now, required disks are always included first; local disks are second, and optional disks selected on the command line (DA=...) are third. Code was also added to ensure that a disk is never included twice. Finally, awk was used in the shell version in an attempt to speed things up a bit. The perl version is still lots faster, but awk does beat the shell looping formerly done. Moved nowhere. 9958. August 4, 1998 FITLD Chris Inserted code to check the value types associated with the global interferometry data interchange keywords and to correct them (with warnings) whenever possible. This should prevent reference frequencies being rounded to single precision. The tolerances used in comparing frequencies have been revised accordingly. Moved nowhere. 9959. August 7, 1998 $QPGOOP/OMFIT.FOR Ketan At user request, allow printing of the covariance matrix by a user-controlled switch. For one component models with ten or less parameters, the full matrix is printed out as a matrix; otherwise, the matrix is printed out one element per line. Trapped an overflow that caused some models to crash even though some of the parameters were being held fixed. Fixed a logic error that confused the maser model to use the wrong spectral profile slowing convergence. Moved nowhere. 9960. August 7, 1998 $APGNOT/DTSIM.FOR Ketan $APLNOT/DTPARM.FOR $INC/DTPM.INC Changed the internals of DTSIM to process station coordinates using XYZ not Longitude, Latitude, and Elevation. Changed the inputs processing routine to accept both types of inputs. Need to relink $APGNOT/DTCHK as well. Moved nowhere. 9961. August 7, 1998 $QPGNOT/KRING.FOR Ketan At user request, added the delay/rate search window sizes to the screen messages when PRTLEV>0 . Moved nowhere. 9962. August 7, 1998 $APLNOT/GEOPOS.FOR Ketan $APGNOT/DTSIM.FOR The subroutine GEOPOS introduced into DTSIM just today is needed by DTPARM and is now to sit in APLNOT. Moved nowhere. 9963. August 7, 1998 AIPSWC Eric Checked in the aips system line counting procedure so that it won't get lost. Moved nowhere. 9964. August 7, 1998 IMAGR.HLP Eric Added Bill Cotton's notice about WWW tools to help set the adverbs to IMAGR for interfering sources. Moved nowhere. 9965. August 10, 1998 $APGOOP/CL2HF.FOR Ketan Merged last version of CL2HF cleaned up in July 1997 with latest version submitted by Dave Gordon in May 1998. Changes include: cleaned up frequency calculations, correlator model now accessed using CL table and not IM table, some re-formatting of comments and some re-formatting of FORMAT statements. Moved nowhere. 9966. August 11, 1998 Misc Eric AU1A Changed test on page full by one line to allow for FOUND IN message AU8 Changed handling of errors in EXTDEST to close the file and to take blank as NO in the call to CONFRM. UVSRT Increased the work buffer from 256 Kilowords to 1280 Kilowords. It should go faster now. Moved nowhere. 9967. August 12, 1998 $QPGNOT/KRING.FOR Ketan Although KRING requires less memory than FRING, its been running dog slow. It turns out that KRING has been trying to solve FFTs 16x larger than FRING. Turning this down to the same level as FRING results in a 15+factor decrease in execution time. The solution extrapolation routine had some logic flaws which have now been fixed. The new SNR calculation is now in place but is not yet the default. Moved nowhere. 9968. August 12, 1998 Cal system Eric Changing gain application code so that the gain records in the CL or SN table do not need to occur at the same times, but in strict time order: Fixed errors handling subarray and FQ selection: DSEL.INC Added NTIMCL, NEXTCL to point at the next CL record to be used and TIMECL(2,MAXANT) to hold the current times. PUVD.INC Added parameter MAXFQ=256 max number FQ IDs. and increased XCTBSZ from 2500 to 5000. This then allows a limit of 1000 for num ant * num pol * num IF which should do (lthough it is < 5040 which is MAXANT*MAXIF*2). DATCAL Minor code cleanup. Still needs new interpolation at the same times as before. CGASET Moved interpolation weight setting inside loop on antenna and use the new antenna-dependent times. CSLGET Fixed handling of SUBARR, FREQID - they are not acceptable as zero although table records are. Changed logic all around to fetch the records needed for interpolation on an antenna basis. SDCGET SUBARR = 0 meaningless - dropped. Table can have 0. BLGET FRQSEL, SUBARR = 0 meaningless - dropped. Tables can have 0 meaning all. BPASET Was ignoring SUBARR entirely, mixing subarrays happily. Added proper tests and fixed FREQID test. BPGET Was ignoring SUBARR entirely, mixing subarrays happily. Added proper tests and fixed FREQID test. UVGET Added test and error exit for SUBARR < 1 or > # of subarrays when calibrating. It sets it to 1 if there is only 1. SNPLT Renames local TIMECL to avoid possible problems. Moved nowhere. 9969. August 12, 1998 GETANT Eric NSTNS was a common parameter widely used in all sorts of places. I did not realize this when I checked antenna usage and so allowed NSTNS to be set to the number of records in the file. Actually this was what was intended by the original designer with TELNO(i) giving the actual antenna number for record i and the other arrays being (j) where j-actual ant number. But, no one else seems to have known of this clever if difficult usgae. Some tasks used this well, others messed up, and the polarization cal application seems to have ignored it entirely. Changed so that TELNO(i) = i always. Moved nowhere. 9970. August 12, 1998 UVPLT Eric Added EIF to the adverbs and fixed it to loop over subarray and loop over IF if requested. Bandwidth synthesis may now be fully plotted. Moved nowhere. 9971. August 12, 1998 INDXR Eric Changed arguments to OGTDEL and OGETMC to R*8 for times. Changed to allow the VLBMOD. Fixed TABLEUTIL and GETMC to use CLTIME as Double precision which it is supposed to be everywhere. Moved nowhere. 9972. August 12, 1998 UVDATA Eric Added keyword DOUVSCAL which when true in a read uv data set will cause UVREAD to scale the input u,v,w by the ratio of the output to the input frequencies. Certain u,v copying routines may need this (e.g. SPLIT, UVCOP, FLGIT do something similar) but others like IMAGR transfer the input frequencies suitably to uv scratch files and do not want this option. Moved nowhere. 9973. August 12, 1998 u,v,w scaling Eric Some routines neglect to scale u,v,w from the input ref frequency to the frequncy being displayed/copied etc. Changed DECOR Scale u,v,w when transfer to output UVPRT Scale displayed u,v,w to BIF, BCHAN UVFND Scale u,v,w displayed and used to channel, IF. CLPLT Scale u,v,w as read for UVGET affects. Also dropped several unused subroutines. FRPLT Changed how frequency scaling is dopne in phase shift, including getting the frequencies with CHNDAT. IBLED Scale u,v,w as read into averaging routine. POSSM Scale u,v,w as read - it already scaled for channel but not for base-band shift VPLOT Scale u,v,w as read for FREQ/UVFREQ. Moved nowhere. 9974. August 12, 1998 EXTLIST Eric The forgotten AU8A was updated for UVPLT - I will bet it does not work well onlots of plot type (esp CLPLT, VPLOT, ..) Moved nowhere. 9975. August 12, 1998 $APLNOT/LISRAN.FOR Leonia The new routine LISRAN returns array of the list of the random parameters and the relevant array which informs whether the given parameter should be averaged (1), summed (0), or not averaged, not summed (-1). Moved nowhere. 9976. August 13, 1998 $APGNOT/UVAVG.FOR Leonia UVAVG has treated correctely the 9 random parameters. If the number of random parameters exceeds 9, a warning has appeared, informing the user that this exceeding can be a reason that AVG MAY BE BAD. This warning confused users, because they has not have enough information to estimate how serious it is. I modified the warning. Using the new routine LISRAN (look 9975) I give the list of the extra random parameters which I think clarifys situation for the users. Moved nowhere. 9977. August 13, 1998 FITLD Chris Removed a test feature that had been left in the cal transfer code. This caused calibration tables loaded from different input files to be written to separate output tables. All calibration data is not concatenated in table version 1. Added a note in the explain file with recommended TAMRG settings for removing any duplicate records from the version 1 tables. Moved nowhere. 9978. August 13, 1998 PHNEG Chris A new task to flip the sign of the visibility phase and contributed by George Moellenbrock. This is needed to fix some data correlated in Mitaka. Moved nowhere. 9979. August 13, 1998 FLGIT Eric Corrected it to recale u,v,w when output frequency is not the same as input. Also changed it to copy the FQ file reformatted including (if multi-source in, single-source out) the source frequency offsets. Changed it to loop over subarray when there is more than 1 and the user asks for all. The fix yesterday caused SUBARR=0 to fail even when there is only one. That fact made me alter UVGET to move the test on time order and subarray above all calls to calibration preparation routines. Now, with 1 subarray, it will work with SUBARRAY=0. Moved nowhere. 9981. August 14, 1998 TI2HA Eric New task to convert times in a UV data set to hour angles (offset to be around 1.0 to avoid negatives). This will allow averaging of multiple days if desired, but prevents further calibration. Moved nowhere. 9982. August 14, 1998 $APGOOP/CL2HF Ketan for Dave Gordon Updated CL2HF's date processing to properly convert the dates into the desired Haystack format. Also unnecessary call to obtain REFPIX was removed. Moved nowhere. 9983. August 17, 1998 LISTR Eric Fixed handling of LIST mode to avoid dividing the FREQ by 1.E9 on each iteration and to loop over IF since it claims to be able to do so. Moved nowhere. 9984. August 17, 1998 POPS MEmory file size Eric Large procedures are capable of blowing core still. Changed DCON.INC To get sizes from an include called PKSZ PKSZ.INC To set K and LISTF sizes (4 times larger roughly) and the legitimate Save/Get version number (now 16). Also set KKT size 3 times larger. AU2A Use PKSZ rather than a data INIT Use PKSZ rather than hard-coded numbers FILAIP Use PKSZ rather than hard-coded number FILAI2 Use PKSZ rather than hard-coded number FILINI Trap ME now and use PKSZ for size SGLAST Reset MPAGE and LPAGE after call to SGLOCA using PKSZ. Changed to invoke GET LASTEXIT when it fails because the version is too old. This uses the page full printing trick to handle type aheads. STORES To use whatever MPAGE and LPAGE come out of SGLOCA, but reset to standard after disk is read. SGLOCA Remove SGVERS setting to PKSZ, add array of former MPAGE and LPAGE sizes for versions 14 and up, set MPAGE or LPAGE to the size appropriate to the outdated file about to be read. POPSGN Use PKSZ to set values into K, KKT array. AIPS Use PKSZ to set KKT parameters. AIPSB Use PKSZ to set KKT parameters. AIPSC Use PKSZ to set KKT parameters. DAIP Use PKSZ to set KKT parameters. Moved nowhere. This will require creating new ME files for everybody, old SG files will survive! 9985. August 17, 1998 FREQID = -1 Eric The pernicious FRQSEL or FREQID is set to -1 when there is no FQ selection - but I dropped that from the cal routines since one cannot mix Subarrays of FQ ids. Allow FRQSEL=-1 again in CSLGET, BLGET, BPGET, BPASET. Moved nowhere. 9986. August 18, 1998 FITLD Chris Corrected the reference frequency for lower sideband data when concatenating data. Moved to patch area on this date. 9987. August 19, 1998 UVCOP, SPLIT Eric Changed both tasks to eliminate the random parameters called REMOVED which eat up space for no good reason. This is tricky - please watch these tasks to make sure I did not miss some case. Moved nowhere. 9988. August 20, 1998 FITLD Chris The calibration transfer routines were looking up the order of the IFs in the output file in the FORDER map using the remapped frequency ID while FORDER was constructed using the unmapped frequency ID. This would cause FITLD to place calibration data in the wrong elements of the output arrays if the frequency IDs changed when a file was appended; since the faulty addressing usually resulted in a zero index, this was most likely to lead to a bus error. Moved nowhere. 9989. August 20, 1998 BLCAL Eric BLCAL fell into the trap of setting NKEY (to 3), then calling TABINI, and then using the output NKEY (now about 50 or so) to call TABKEY. Fixed it to reset NKEY to the desired number - all other $APLNOT *INI.FOR have this right by now. Moved nowhere. 9990. August 20, 1998 IMAGR, MX Eric Both tasks refused to handle 1-pixel windows. I changed QCLEAN to allow single pixels to go to the Cleaning if there is a single-pixel window (among the clean boxes), otherwise single pixel fields are skipped. In MX, I changed it to go ahead and Clean any single-pixel field. I also fixed the TV scaling which used the max/min in Clean boxes which is disasterous for a single-pixel box. Moved nowhere. 9991. August 20, 1998 APCAL Chris APCAL could only handle a maximum of 10 gain-curve records. Increased this limit to 2*MAXANT. Moved nowhere. 9992. August 21, 1998 BLINI Eric I forgot to relink everything for yesterday's fix of BLINI. Moved nowhere. 9993. August 21, 1998 UVCOP Eric Corrected 2 bugs: a pointer was updated while the old value was still needed and the output gain for compression was misaddressed. The copies of compressed data were therfore invalid and things could blow up or do odd things to the source table. Added SUBARRAY as an option. This means testing for it, resetting the time and subarray when it is used and copying only the correct antenna file. Also it meant fixing MCSEL, IMSEL, and PCSEL to have and use ISUB and JSUB options of the other selection subroutines and to take real advantage of these options by setting them properly for SUBARRAY > 0 (and also BPARM(2) > 0). Moved nowhere. 9994. August 22, 1998 OERROR Eric Changed code to leave the COMPRESS RUN file in $RUNFIL on error termination. Added messages telling user how to fix things. Moved nowhere (tested on pilabo). 9995. August 22-23, 1998 CookBook Eric Changed things for EDITR, EDITA, SCMAP, SNEDT changes: COOK6 Remark about number of baselines COOK9 Remark about number of baselines COOK10 Reword the section on EDITR COOK5 Rewrite the section on EDITR. COOK4 Adjust the wording on EDITA, replace figure EDITA.PLT. COOKI Adjust things a little for 4 and 5 changes COOK0 recipe gone from 5 RECIPE recipe no longer used Moved nowhere. 9996. August 22-23, 1998 EDITR et al. Eric Changed QEDIUTIL to be smarter about auto-correlation only data sets. No SHOW ALSO menu item is needed and SHOW PHASE is suppressed. Also fixed it to update the screen fully when a plot is pending and SELECT FRAME is chosen. Changed it to make the lower (edit) plot a bit bigger than the others. Moved nowhere. 9997. August 24, 1998 UVCOP Eric Did not set the source number until there was a success, but the CURSOU was referenced in flagging. A delete a source for all time could delete all data for all time thereby. Also found yet another reference to IPTRO after it was already incremented - so moved the increment code yet again. Moved nowhere. 9998. August 25, 1998 $YPGNOT/CONFI.FOR Leonia For two circle constraint, when the inner circle of the bigger configuration is used as outer circle of smaller configuration it is usefull using the same pads for the common elements of the common circles. For this purpose IN2FILE is used now to store the position of the fixed elements at the outer circle of the given configuration. During optimization these elements are not moved. The .HLP file is updated. Moved nowhere. 9999. August 25, 1998 VPLOT Leonia Historically VPLOT has not labeled vertical axis if weight is plotted. I added the labeling. If NCOUNT is selected very big erronously, VPLOT has failud to finish succesfully without warning what happened. I added the maximum of plots on a page: 40. If NCOUNT >40 the task stops with explanation the reason. Moved nowhere. 10000. August 25, 1998 $APGNOT/SPLAT.FOR Leonia I can not wait more to put the new task in the system, because I can not wait the next even ten thousand. The new task SPLAT having all ability of SPLIT, provides additional options: 1. It can assemble selected sources having calibrated and edited (by FG table) into new multi source file. 2. The averaging in time and averaging subsequent group of frequency channels is added. The task can be usefull when the preliminary calibration which reduced fringe rate and delay carried out. The output of SPLAT will have much less data because the averaging. Moved nowhere. 10001. August 26, 1998 $QPGNOT/KRING Ketan My change of August 12, 1998 was ill-considered. KRING was only running slowly because it was actually able to handle the heavily padded FFT that FRING choked on. Turning down the padding level made KRING's FFT search unacceptably inaccurate for the Least Squares. Ive now turned the padding level back up to what it was before and introduced a new handle, CPARM(10), that allows the user to turn down the padding manually. CPARM(10) defaults to a factor of 8 in delay and rate, and can be set to any positive power of 2. KRING will not allow values less than 1 and will, in all cases, only use a padding level consistent with the amount of available memory. Moved nowhere. 10002. August 27, 1998 TABKEY anticipation Eric Changed TABKEY using routines to check floats for type and read the appropriate one into the variables being returns. Then 15OCT98 will be able to read tables made by 15APR99 (in which all will be real*8). Changed: ANTINI Add lots of tests for current R*4's. Note that the REAL*8's are gotten by RCOPies since alignment in the data buffer is absolutely not guaranteed. ACTFRQ Proper equivalence and test for GMMOD. BPINI Added equivalence, stop copying REAL*4, add tests and correct handle R*8 if it is returned (not expected) CALINI Add REAL*8 test and handle for GMMOD. GAINI Add REAL*8 test and handle for GMMOD and offsets GAININ Add REAL*8 test and handle for GMMOD. GPINI Add REAL*8 tests, handle for various parameters, increase buffer size, fix error the Height was not set and the lat was set to the height. IMREF1 Like IMIO below. MCINI Changed call seq - freq should be REAL*8. Fixed to have FREQ REAL*8 and to handle either on read. MCSEL Changed call seq to MCINI. PDRGET Correct message suppression handling. Changed to read either R*4 or R*8 values. SNINI Changed to use simpler equivalence scheme and to handle reading REAL*8's TABKEY Clarified a comment or two, CLCAL Changed a few comments around (it writes an R*4 only). FITLD Changed call seq to MCINI. Changed to handle a table key read by specifying the needed keywords rather than reading "all" limited to 50. FXPOL Changed FIMIO like the routines below. MFPRT Changed TABKEY ALL read into a proper read to get the DEPTH parameter which was being set to 1 otherwise. UNCAL Added R*8 test for GMMOD. SNPLT Added R*8 test for GMMOD. There is a collection of FITLD routines with real errors in them which needed fixing. They were doing RCOPY REAL*8 variables onto REAL*4's in a common and were upping counters by 2 for REAL*8 not NWDPDP. The include files also need their buffer for keywords increased for possible REAL*8's. Changed DATV.INC DBAV.INC DBCV.INC DCAV.INC DCMV.INC DFRV.INC DIMV.INV ATIO BAIO BCIO CAIO CMIO FRIO IMIO AGIO SOIO (includes okay) Moved nowhere. 10003. August 27, 1998 FITLD Chris Generalized the processing of random parameters in UV_DATA tables from VLBA-style FITS files by introducing a table of parameters to be deleted and a table of parameter translations. FITLD now handles correct u, v, w label patterns (UU, UU--SIN, UU--NCP) as well as the incorrect UU-L pattern used by the VLBA and appends the projection suffixes in the output file (UU, UU-L, and UU--SIN become UU-L-SIN in AIPS while UU--NCP become UU-L-NCP). Before this change FITLD would only recognize parameters labelled using the UU-L pattern. NOTE: UU-L coordinates are assumed to be in seconds in UV_DATA tables and not wavelengths as the name suggests so that we can still read broken UV_DATA tables from the VLBA archives. Moved nowhere. 10004. August 28, 1998 $YPGNOT/CONFI.FOR Leonia Opimization with some elements fixed at the outer circle was added recentely for the two circle configuration. (see 9998). Now I added this option for the three circle configuration. The new part of task responsible for reading the fixed positions has been formed in separate subroutine. The HELP file has been updated. Moved nowhere. 10005. August 29, 1998 CALIB Eric Added messages talking about self-cal and non-self-cal modes at the decision points. Users were getting non self-cal quitely when they expected self-cal but selected > 1 source (say with QUAL). Moved nowhere. 10006. August 31, 1998 DATCAL in APLNOT Chris Recent changes to the calibration routines changed their interface so that CALTIM(3) no longer represented the time for which the calibrations were calculated. This change was not propagated to the rate correction in DATCAL which calculated phase changes with repect to the new definition of CALTIM(3) leading to an incorrect rate correction that was approximately twice what it should have been. Replaced CALTIM(3) with LCALTM in this expression. Moved nowhere. 10007. September 1, 1998 CGASET in APLNOT Chris CGASET set LCALTM to a large number after exhausting the calibration table. This made a mess of the rate corrections (see the previous change) if data continued past the end of the calibration table. Since the sole purpose of setting LCALTM to a special value appears to have been the avoidance of redundant recalculations of the complex gains and since this avoidance should not normally be a significant optimization, LCALTM is no longer set to a magic value when the calibration table is exhausted. Moved nowhere. 10008. September 1, 1998 EDIT class Eric Changed: QEDIUTIL 1. FLAG FAST and POINT ignored the all antenna flag. 2. It just averaged subarrays with no thought. Now it forces a single subarray. 3. It forced OUT_FGVER to 1 when it was 0 and used OUT_FGVER for input FG table for SN and TY modes. 4. If an output FG table is written, pick up its version and store it back wih the input, so if there is a loop the same output FG will then be used. 5. Also for SCMAP, get the FG table for the UV data and if it is none, change it to the new FG table. EDITA Changed it to allow in and out FG versions to be different using the same logic as EDITR. EDITA.HLP Ditto UVUTIL Changed UVCOPY to loop over subrrays - can't cal more than one at a time. INPUT Changed several declarations of input character variables not to be specific about length. Otherwise keywords were required to be 8 characters on input and 32 on output. SCMAP has bugs still and so was not putbck yet of COMLNK'd. Moved nowhere. 10009. September 3, 1998 BooBoos Eric Did not edit the TABKEY advance changes properly for ANTINI and MCINI. The former did not matter much, but the latter caused an abort because it used the previous type (double) for what is actually a REAL*4 leading to a 10**43 value. Moved nowhere. 10010. September 4, 1998 FRING Eric Changed it to look for frequency increments only in one IF if the IFs are being done separately and to check the frequency increment and fix it if it is < 0.1 of a normal channel increment. Fixed up some the the ABS value handling of frequency differences as well. Multiple IFs from double sideband were having two channels overlap at 0 frequency difference. Moved from MPI this date. 10011. September 4, 1998 EDIT class Eric Changed: FCINI A buffer was too small - the HPs were the ones that had problems with essential pointers being overridden. REFFAZ Changed it to fix phases even if weight <= 0 and to fix refant even if sol'n = INDE. Also changed it to change the reference antenna when the current refant=0 if the solution is INDE (which it should be in that case). QCLEAN Correct TV loading to check for TXINC wrt the windows. Failures were caused by truncation/rounding in CLBCHK. SCMAP Changed TV menu to select Subarray to edit. Changed it to make a cataloged output file of the phase-only calibrated data when switching to phase and amp. It cannot use the output file for this when there is more than one subarray because one cannot copy over the data set applying cal with > 1. Fixed display of rms. UVUTIL Copy tables only once. Changed UVCOPY to loop over subarrays calling UVCOP1 (the former UVCOPY). Moved nowhere. 10012. September 4, 1998 DATCAL Chris JNDXA1 and JNDXA2 were not incremented in loop 25 so that DATCAL only checked for non-zero delay corrections in the first selected IF (EIF - BIF - 1 times). This resulted in delay corrections not being applied in cases where the delay corrections were zero for the first selected IF but not for others (usually as a result of missing data). Moved nowhere. 10013. September 6, 1998 $APLNOT/DATBND Ketan When DATBND and BPASS were modified on July 16, 1998, CPASS was forgotten. As a result, bandpass corrections generated by CPASS were incorrectly applied for AC data in DATBND. This had no effect upon cross-power spectra. Moved nowhere. 10014. September 6, 1998 $APGOOP/CL2HF Ketan I was supposed to begin versioning CL2HF as of Aug 11 but forgot to. It begins now. The date format change of Aug 14 is the first entry. Also, the versioning is to be mirrored in the help file. Moved nowhere. 10015. September 7, 1998 $QPGOOP/OMFIT Ketan At user request, changed user input for early termination from fractional change to percent change. Fixed bug in error-bar calculation that caused error-bars to be sqrt(2) too small. Moved nowhere. 10016. September 7, 1998 $APGNOT/DTSIM Ketan Change noise calculation in DTSIM so that with a 0.5 MHz bandwidth, 2 sec integration time, and SEFD of 295 Jy [inclusive of all efficiency losses], results in a noise level of 0.295 Jy/vis. Moved nowhere. 10017. September 8, 1998 FRING Eric Added code to cause the frequency increment error message to come out only once per subroutine. Moved nowhere. 10018. September 8, 1998 QCLEAN Eric The dubious option of quitting on the first negative clean component was supported in the inner loop, but no where else. Changed things to get the actual residual value, to preserve it if negative even when the re-imaged fields have a positive peak, and to set FINISH true on the basis of it. The middle point is why I do not like the option - the peak in the residual was positive after a re-imaging even when it had last found a negative component. The first negative is not a meaningful concept. Anyway, the routine now does as advertised. Moved nowhere. 10019. September 8, 1998 LPSEC in APLNOT Chris Added the upcoming leap second for December 31 1998. Moved nowhere. 10020. September 8, 1998 GPSDL Chris Added voluminous diagnostics at PRTLEV 2. Also added a missing call to OGPTAB in TECMOD. Moved nowhere. 10021. September 9, 1998 CLCAL, SN2CL Eric Put the force subbaray to 0 option back into CLCAL including subroutine SN2CL and the help file. With the new way of applying cal's, now debugged I hope, this should work. Moved nowhere. 10022. September 9, 1998 $APGOOP/CCEDT.FOR(.HLP) Leonia Craig Walker asked to add an option of cutting components. Now for each component, all components within +-CPARM(3) in X and +-CPARM(4) in Y (in arcseconds) are summed. The flux density test (CUTOFF) is then applied to the summed flux density. The HELP file has been updated. Moved nowhere. 10023. September 10, 1998 $QPGOOP/OMFIT.FOR Ketan Fixed user-encountered bugs with the simple self-cal mode. Added hooks for fitting multiple sources simultaneously. It is now no longer necessary to specify the source name when ultiple sources have not been selected. At user- request, added option for writing output data set with only model instead of data-model. All the model-types in OMFIT are automatically now available for generating fake data. This reproduces all the functionality of UVMOD except noise generation, weight modification, and the exponential disk [EXPD] type (ive never had a request for that one...). OMFIT additionally allows an arbitrary number of model components to be added simultaneously instead of one at a time. OMFIT's 'MAS' and 'ZEE' model-types allow components with spectral profiles. The 'POL' and 'ZEE' model-types allow components with polarization properties. An almost-complete test suite of simulations is available to test that OMFIT does produce the correct results for each model-type and for the self-cal modes. Enough already... Moved nowhere. 10024. September 11, 1998 Fitting Eric IMFIT, JMFIT, and SAD had trouble with images that have no Clean Beam parameters. They put out 0.0 for error estimates under that circumstance. Fixed them not to do so. Also fixed math routines in SAD that seemed able to try to take EXP{+128}. Changed IMFIT and JMFIT to get parameters in a normal way. Getting them with an array with indicies that have to change all the time is bad. Added option to use a beam-smeared Clean Beam as the initial guess and the Clean Beam for deconvolution. The user must provide the ratio of the bandwidth to the center frequency which will be used to scale the FWHM of a Gaussian smearing function. Wrote RECONV, a routine to convolve two Gaussians, and added BWSMEAR to POPSDAT and NEWPARMS. Moved nowhere. 10025. September 16, 1998 FRQUPD Eric Used the wrong format number causing it to blow up on error. Also repeated formats... Moved nowhere. *********************************************************************** 15OCT98 as NEW *********************************************************************** 10026. September 20, 1998 $QPGOOP/OMFIT Ketan Indexing error caused SN table time-tags to be off by one solution interval when self-calibrating data. Moved from TST tomorrow. 10027. September 21, 1998 $QPGOOP/OMFIT and HLP Ketan Indexing error caused covariances between self-calibration parameters to be ignored - slowing convergence when self-calibrating data. Also loosened convergence criteria until self-calibration error-bars are properly sorted out. The weights reported in the SN table may not yet be correct. Moved from TST this date. 10028. September 21, 1998 SPLIT Eric Changed INDXIN To return -1 when there is no error except that no matching index record is found UVGET To handle the -1 return from INDXIN SPLIT To handle -1 returns from UVGET better (keep going if looping over subarray, else go on to next source). Also to not send subarray 0 into UVGET at all. Changed default for NPOINTS to all channels. Fixed count of channels averaged to apply to FQ table values. SPLIT Fixed help for new NPOINTS default. DATCAL Changed logic counting samples so that those either previously flagged or flagged by gain would be known as dead. Moved to 15OCT98 this date. 10029. September 21, 1998 TVFLG Eric There were several arrays of 5000 meant to handle #subarrays * #baselines. Changed these to 15000 for Frazer. Moved to AOC today in SUL and to 15OCT98 this date. 10030. September 22, 1998 $QPGOOP/OMFIT Ketan Strange typo introduced yesterday is now fixed. Also, found an array out-of-bounds condition that was only triggered for large self-calibration problems. Moved from TST this date. 10031. September 22, 1998 FITLD Eric Added comments in DOCONCAT section about setting OUTSEQ=0 for concatenating. Moved to 15OCT98 this date. 10032. September 22, 1998 UVCOP ERic Selecting by source in the CL copying can cause problems with CL tables that are not set up well wrt the sources being copied. Changed it to call CLSEL with 0 for the number of selected source (which means no selection applied). Moved to 15OCT98 this date. 10033. September 30, 1998 $QPGNOT/KRING.FOR Ketan Fixed a typo that prevented multi-band delay solutions from being transferred to all IFs in the solution table. This only affected the 'SMBD' and 'MBD' OPTYPES. Moved from 15APR99 this date. 10034. September 30, 1998 $HLPFIL/SAD.HLP Ketan At user request, added missing description of DPARM(6) to inputs section of help file. Moved from 15APR99 this date. 10035. September 30, 1998 $QPGOOP/OMFIT.FOR Ketan Squashed some bugs, added some new ones. Fixed SN table writing routine. I was writing out each solution twice with different times. Also, fixed the writing of GAINS model info to the SN table so that GAINS model solutions can be used with other AIPS tasks via the SN table. At user request, I added a new model, the BLOB model. This model allows for a Gaussian component with variable position. Currently, the only linear variability with time is implemented. This could easily be generalized to non-linear variations with time, frequency, channel, IF, etc. [any requests?] Moved from 15APR99 this date. 10036. October 8, 1998 $APLAXLIN/various Pat Placed the APLLINUX versions of ZGRLHI, ZGRLTC, and ZPRMPT in APLAXLIN to test them with the new readline v2.2 library. Readline now works under this architecture too. Moved from 15APR99 this date. 10037. October 8, 1998 $AIPSPUBL/AIPSMEM101.TEX Ketan $AIPSFIG/AM101[A,B].EPS Fringe-fitting memo is now checked in. 10038. October 8, 1998 $QPGNOT/KRING.FOR Ketan $HLPFIL/KRING.HLP New SNR calculation has been moved into place in KRING. The HLP file now reflects the new calculation, proper references to the Schwab and Cotton paper, AIPS MEMO 101, and fringe-fitting now in explain section of HLP file. Moved from 15APR99 this date. 10039. October 16, 1998 $QPGNOT/CVEL Leonia CVEL has not worked in 15OCT98 and 15APR99 if SUBARRAY=0 is used. SUBARRAY has not been recalculated to 1 as it is written in HELP. I added this recalculation. Moved from 15APR99 this date by Pat Murphy.