[daip] md5sum, AFS

[Eric Greisen]

Miguel de Val Borro writes:

> I've connected to urania.aoc.nrao.ed via FTP and downloaded the
> files pub/software/aips/31DEC02/31DEC02.tar.gz, 31DEC02.tar.gz.md5sum
> and install.pl using binary transfer.
> The output from 'md5sum 31DEC02.tar.gz.md5sum'
> (c3b0dbe4b023b9a211cb3beee5eccabb  31DEC02.tar.gz.md5sum) is different
> from the contents of the file 31DEC02.tar.gz.md5sum
> (c6a3eecf851244962dcf0a4cc997d580  31DEC02.tar.gz). I tried and
> repeated the download a few times today November 24 with the same
> result all the time. Could you please give me a hint? Is this step
> necessary to perform the installation?

    You need to run md5sum on the tar ball not the md5sum file.  The
correct command is
       'md5sum 31DEC02.tar.gz'
This is to test whether you recveived the tar ball correctly.  If you
did not the install will fail so if the md5sum does not check
correctly you may proceed but may encounter problems.

> 
> Another question is that we are doing the AIPS installation in an
> AFS file system; do you know of any possible difficulty in this case
> because of the installation scripts may be using NFS syntax? 

So far as I know, we do not do anything that is "NFS syntax".  The
package assumes that environment variables translate into pathnames
that lead to files on disks that are mounted to the machine running
things.  It also assumes that sockets may be created and used.  I do
not think that the new AFS should break anything, but, of course, one
does not know without trying.

Eric Greisen