IMAGR - cleaning multiple fields ...

[Eric Greisen]

R. Niruj Mohan writes:
 > 
 > > Now - what size have you set for the AP memory?  I need to see
 > > $INC/NOTST/DEC/PAPC.INC and also the output of RUN SETPAR where you
 > > ask to change parameters (code 2) - it shows the size you have
 > 
 > Surprisingly, the DEC area is empty ! I found PAPC.INC files in the
 > directories ALLN, CRI, CVEX and IBM but $ARCH is ALPHA !

     We often do not ship code with large APs pre-compiled even if we
are using them locally.

 > 
 > SETPAR has the foll relevant statements :
 > 
 >  26  No AP batch starts: weekend hours  15.00016.500
 >      No AP batch starts: weekday hours  13.00017.000
 >  27  AP roll interval minutes            5.000
 >      AP patience: (1)*N+(2)*N*N minutes  5.000 1.000
 >  28  Line printer width (72 - 132)         132
 >  29  Pseudo-AP 2nd memory (1024s)         1216
 >  30  Max length of "short" vector            0

     This is the standard size so it is ok.

 > 
 > 
 > > There is a bug in the code casuing the UVPGET failure reported.  I
 > > think that you can get around it by putting the largest field first in
 > > the list of fields.  What IMSIZE are you using for field 10 at the
 > > moment - it was the one that failed.
 > 
 > You are right ! There was a problem in compiling QCLEAN.FOR on our machine
 > so I had ftped the binaries and copied them over to the correct areas. I
 > had assumed that the ftped binaries had been corrected for the patches - I
 > didnt know they were not. The 10th field is indeed larger than the ones
 > before it. I ran IMAGR again with the fields in descending order of
 > fldsize and it seems to be running ok (it is still cleaning but it looks
 > like it is fine!). I shall get the patches and COMPRL and COMLNK the
 > relevant tasks. 

       What version are you running?  You explicitly referred to
15APR99 in your last mail.  The 15OCT99 patches are not applied to
(and should not be applied to) 15APR99 versions - even if the bugs
existed back then too.  The patch for the bug you encountered was
applied to our local version of IMAGR.EXE in December.  That is not
the ftp area version however which we do NOT change once frozen.

 > 
 > Another problem with the 15OCT99 version. When I run TVMOVIE, it displays
 > all the images in an array form but when it supposed to start the actual
 > movie, it exits saying there is a 'TV problem'. Can you help me out with
 > this as well ?

     Again what version?  And I do not know what to say with only this
amount of inforrmation.  TVMOVIE works fine for me.  What are your
inputs?

 > 
 > A query : Since QCLEAN.FOR would not compile, and we started getting more
 > and more tasks failing (MX, IMAGR etc etc - different users here use
 > different tasks), I just ftped the binaries and copied all the binaries to
 > the $LOAD area. Everything else was configured when I installed aips
 > afresh from the source code. I also copied over the ftped library files.
 > Is this alright ? Or did I screw up things by doing this ? In which case,
 > I shall have to ask for a binary version on a CD-ROM or exabyte (in which
 > case, is the 15DEC99 version more bug-free than 15OCT99 version ?).

       I do not understand your assertion that 15OCT99 is particularly
buggy.  A version under constant development will always be buggier -
and more fixable - than frozen versions.  BUT you must have a compiler
that works or you cannot consider the 31DEC99 version at all.  We do
not distribute binary copies of it.

Eric