Re: RFS: libpgjava
On 2012-07-27 02:13, Andrew Ross wrote:
> On 26/07/12 06:20, Niels Thykier wrote:
> Hi Niels,
> I think it's at least worth fixing bug #659324 which is fixed by
> adding in transitional package libpg-java, since this fixes the
> upgrade path for this package. If you prefer I could prepare an
> upload that only contains that change.
Certainly, the transition package should not pose a problem (other
than it has to go through new?).
> Annoyingly since the last release upstream have moved from SVN to
> git and removed all the svn tags from their headers, so the change
> affects the comments in every file. However the changes from  of
> any real consequence are:  which fixes upstream bug 6293 with
> the following comment:
> This solves a major performance problem for ResultSetMetaData
> users which did not cache the ResultSetMetaData object. One of the
> users is the driver's own implementation of updatable ResultSets,
> so this can't be worked around solely in end user code.
> and:  which corrects a signed/unsigned issue
Will the reverse dependencies handle these changes correctly?
Technically  is an API/ABI changes for libraries extending the
AbstractJdbc2ResultSet class (but if nothing extends it directly, it
won't be a problem I guess).
Regarding , do clients/rdeps handle negative values correctly (or
are they not exposeed to it)?
(For reference, dak finds the following reverse dependencies
Checking reverse dependencies...
# Broken Depends:
# Broken Build-Depends:
> They look fairly safe to me, but I'm happy to go with an update to
> the current upstream version if you think that's better.
> Thanks, Andy
>  https://github.com/pgjdbc/pgjdbc/commits/REL9_1_902 
I am fairly certain a transitional package will be accepted, so I am
okay with just that. For the others, I would like the release team to
pre-approve it before the upload. (i.e. create a full debdiff and
send it as an attachment to a new "unblock" bug against