[daip] MAXIF

[Eric Greisen]

Olaf Wucknitz wrote:
> Hi Eric,
> 
> I would like to give it a try and compile a private version of AIPS with a 
> higher MAXIF (and other necessary modifications) to be able to load my 
> current data. I know how to build private versions of tasks (currently 
> using the Intel compilers) and how to install a source version of AIPS, 
> even though I have not done the latter for a while.
> So I think I know how I could set up a patched AIPS in its own AIPS_ROOT. 
> However, what I would really prefer is having this version using the 
> normal configuration and act as an alternative to 31DEC10, 31DEC09 etc.
> In that way I don't have to define data areas and all the other tiny 
> things again and can more easily switch between the versions.
> 
> Could you give me some advice how to set it up in this way? Of course I 
> won't complain if anything breaks. Is this described somewhere?

Simple suggestion - rename 31DEC10 to some name such SAVE10.  Do your 
text install.  Then rename the new 31DEC10 to e.g. 32DEC10 and rename 
SAVE10 to 31DEC10.  In the AIPSPATH.CSH and >SH files in your $AIPS_ROOT 
area there are 4 "standard" version names.  Change CVX to point at your 
test version and then say version='cvx' whenever you need the large 
MAXIF version.  This does not require the help files to be in the load 
area or anything annoying like that (but required for small local test 
tasks).

The main issue would be MNJs.  You could of course do an MNJ on 31DEC10, 
then do 2 renames to make your version 31DE10 and run the MNJ again and 
then rename the directories back.  That will work except if I have to 
change PUVD.INC.  I do avoid that since the cost in rebuilds is very 
high, but it does happen.

I attach a script I use to examine consequences of changes like this - 
run it on your $TST and $CVX versions directing the output to different 
files and then compare.  You could use the result to argue that I should 
increase the public MAXIF.  I may have to increase MAXCIF for the EVLA
but would rather not do MAXIF unless it is cheap.  (New code tends to 
use dynamic memory more esp in the case of number IFs - they are usually 
2 or 4 so 90 is a gross overkill.)

Eric Greisen