Daily snapshots hosed (was Re: [pgadmin-hackers] What about CREATE OR REPLACE FUNCTION?)

Dave Page <dpage(at)vale-housing(dot)co(dot)uk> writes:
> ... I can't find an up-to-date snapshot

> I tried postgresql.rmplc.co.uk and got one (apparently) dated 7 Oct, however
> CREATE OR REPLACE FUNCTION didn't seem to be there (it certainly doesn't
> work anyway - syntax error at OR). I then looked in the primary copy on
> mail.postgresql.org and found the copy there was dated 30 Sept from which I
> assumed that the 07/10/2001 date on rm's copy was actually a US date - that
> site has been seriously out of date before.

I just downloaded
ftp://ftp.us.postgresql.org/dev/postgresql-snapshot.tar.gz
which has a date of yesterday in the FTP archives, but actually
contains a snapshot from around 15 September as near as I can tell.
Looks like something is hosed in the snapshot preparation process;
Marc, could you take a look at it?

>> and I don't know the
>> magic that has to be worked on the PostgreSQL CVS version of the
>> configure script in order to make it run without barfing.

> I always assumed that something is done when the tarballs are built as the
> work just fine on the same machine.

No, the tarballs should be the same as what you get from a CVS pull
of the same date (other than not having a lot of /CVS subdirectories).
In fact, they're made basically by tar'ing up a CVS checkout. Please
try diffing configure from a tarball against one from CVS to see if you
can figure out what's getting munged during your CVS pull.

> The only odd thing I can think of is
> that my copy of the source is maintained on my PC using WinCVS and was
> zipped/ftp'd onto a test box.

LF vs CR/LF newlines leap to mind as a likely source of trouble...
though I'm not sure why that would manifest in just this way...

regards, tom lane