[rt-users] RTFM and Postgres 7.3.

[Patrick Hurley]

Some further comments on the issue/logs I posted here:

http://lists.fsck.com/pipermail/rt-users/2004-January/019901.html

Note that everything I say below are assumptions, as I'm not (yet)
well-versed in either Perl or Postgres -- I'm very open to being
corrected if I'm wrong.

My assumption is that the pg_atoi zero-length string errors that occur
with RTFM are due to the empty ObjectId's shown in the logs for
RT::FM::Class and RT::FM::Article.

Meaning, in the logs linked to above, the lengthy queries during with
the error occurs contain:

(ACL.ObjectType = 'RT::FM::Class' AND ACL.ObjectId = '') OR 
(ACL.ObjectType = 'RT::FM::Article' AND ACL.ObjectId = '')

Both ObjectId's are of type int(11), but have values of '' during the
query, which, I believe, is no longer supported by postgres as of
version 7.3.

Looking at RT::FM::Class.pm and RT::FM::Article.pm, the default value
for id in both cases is ''.  

My assumption is that this could be corrected by changing the default id
value to 0 (or perhaps some other more proper number) for both Class and
Article.  However, I see that both of these files are autogenerated by
SearchBuilder...

Can anyone advise me on:

1) Whether or not my assumptions are correct.

And/or:

2) Assuming my assumptions are correct, where is the place to fix this?


Thanks again,

Patrick