[daip] AIPS in Cambridge

[gtaylor]

Hi Eric,

   After recompiling with 2.95.3 CALIB works properly.  I suggest
a short note on the AIPS web page mentioning that the 3.3.3
compiler joins the long list of those not to be trusted.  This
is a fairly subtle problem that could cause real grief for 
novices.    

   Ciao,
                                              - Greg  

On Thu, 9 Sep 2004, Eric Greisen wrote:

> gtaylor writes:
>  > Hi Eric,
>  > 
>  >    Thanks for the tips.  After a little more thrashing about we fixed the
>  > permission problems and definitions of DA00 and DA01 that were stopping us
>  > and got AIPS to the point where it seemed to be working ok.  We could
>  > create files, etc.  Next we started to use VLARUN to process some data.  
>  > This showed a curious error which is that CALIB finds solutions for only 1
>  > IF.  We tried another dataset and got the same result.  Then we processed
>  > the same data on my laptop and had no problems.  So, my question is have
>  > you ever heard of CALIB failing on half the solutions due to a
>  > configuration problem?  We are using the 3.3.3 GNU compiler.  We are
>  > strongly considering rebuilding aips with 2.95.3.  Ciao,
>  >                                                  - Greg
>  > 
> 
>     Yes I have - with the 3.3.3 compiler.  Itis odd that it flagged
> IFs 3 and 4 of 4 for that user.
> 
>     2.95.3 will work.
> 
> Eric
> 
>  > On Wed, 8 Sep 2004, Eric Greisen wrote:
>  > 
>  > > The $DATA_ROOT definition is never actually used for anything - you
>  > > could just comment it out.
>  > > 
>  > > What I do for most systems is use the $AIPS_ROOT/DATA area.
>  > > 
>  > > In it I make link files named host_n where host is the host name (in
>  > > upper case) and n is 1 2 3 4 5 as needed.
>  > > 
>  > > Thus
>  > > 
>  > > cd $AIPS_ROOT/DATA
>  > > 
>  > > ln -s /home/primate/AIPS/DATA   PRIMATE_1
>  > > ln -s /home/primate2/AIPS/DATA  PRIMATE_2
>  > > 
>  > > and then in $NET0/DADEVS.LIST and NETSP I list the areas as
>  > > 
>  > > /home/AIPS/DATA/PRIMATE_1
>  > > /home/AIPS/DATA/PRIMATE_2
>  > > 
>  > > (where $AIPS_ROOT is called /home/AIPS above).
>  > > 
>  > > I think this is simpler than what we do here which is system-level
>  > > definitions such as /DATA/PRIMATE_1 to point at the actual areas.
>  > > 
>  > > 
>  > > The DA00 areas are also significant
>  > > 
>  > > cd $AIPS_ROOT/DA00
>  > > ln -s /home/primate/AIPS/DA00   PRIMATE
>  > > 
>  > > The proc SYSETUP then populates these.  Note that having the system
>  > > files actually on the host helps with bugs in file locking - which
>  > > does not work reliably across NFS.
>  > > 
>  > > Eric Greisen
>  > > 
>