[rt-users] Re: Postgres "invalid input syntax for integer" errors

[Jesse Vincent]

> Has the invalid value ("") for integer fields on postgres issue been
> resolved in 3.2 series (or in a CVS version)?

I know the behaviour has been improved. I expect to get 3.2.3RC1 out
"soon" and I'd love to hear if that resolves the issue for it.

> The last time I saw such an error in RT itself was when creating a
> queue, it occurred if the priority or other integer fields of the Queue
> were left as the defaults.

That's the sort of thing we like to see in bug reports. Thanks!

> I've just installed the rt3-statistics package from contrib and have
> encountered more such messages, this time they seem to stop the stats
> pages from being displayed.

The stats package doesn't go through the RT API for everything if I
recall correctly. That means that it doesn't take advantage of any
bulletproofing we do against such probelms. MySQL (And possibly earlier
versions of Pg) would automatically convert an integer field of "" to
"0". Current versions of postgres are very strict about the SQL standard
and throw errors to get you to fix your mistakes, rather than
interpreting what you mean and just running with it.   

The failed acl check does fail to _deny_, so this isn't a security issue
per se, but one that I would like to fix if 3.2.3 doesn't already.

Jesse