hey stephen, On Thu, May 25, 2006 at 02:08:13PM +0100, Stephen Gran wrote: > I know that this issue was raised in other reports, but I wanted a bug > just about this issue, since it is fairly straight forward to fix and is > less broad than the other 'is broken' reports.
okay
> I think the correct approach is to first add the user to postgresql as
> the postgres superuser. Then try to connect to the database as that user.
> If that succeeds, then either the package has done the correct thing and
> added the user already, or postgres has been configured to use passwords,
> and we don't need to care.
in the past 24 hours i've been making some svn commits (not quite working
atm) that fixed the problem, but in a different way; the packager could
provide the default authentication method as a variable in the config
script). however, i like your suggestion better because it makes things
more likely to Just Work. so i'd be inclined to do this either instead
of or in addition to what's currently done.
> Hope that's helpful - if you want specific patches, let me know and I'll
> take a look at things. Right now, I've only managed to find the chown
> call, so pointers would be nice.
i don't think it should be too complicated. througout all of the
internal/pgsql shell library, the username used for chown/su are
determined by the _dbc_psql_local_username function... so we'd merely
have to modify that function to check for the user with id as you suggest.
i'll try this out when i fix the current breakage in trunk, if you'd like
to look into this in the meantime feel free to let me know how it goes :)
sean
--
signature.asc
Description: Digital signature
