Hi John,
On Wed, 2016-08-17 at 13:52 -0400, John Harvey wrote:
>
> I had a question about the GDAL package naming convention.
> I noticed that PGDG will sometimes change the name of some packages in
> order to include the version of postgres that they were compiled against
> (or maybe that they require at runtime). An example would be postgis2_95.
Right.
> For reference, in the EPEL repo, PostGIS does not have the 95 modifier.
> Here's a "yum list" result for postgis:
> postgis.x86_64 2.0.7-1.el7 epel
Right, they support only one version.
> I wanted to ask if there's a reason that GDAL didn't follow this
> convention. Even though each version of GDAL has a BuildRequires line that
> specifies a pgmajorversion build dependency, the result isn't a gdal95 RPM.
>
> Is the reason because there's no runtime dependency on postgres (just a
> build one only)?
Well, I thought about adding suffix to gdal before, but then it might break
other packages that depend on versionless gdal.
Noone actually complained about it so far, so I *assumed* it is working. :-)
> Also, if I used a version of gdal that was compiled with PostgreSQL 9.2
> with my postgis2_95 build, would there be cause for concern? With the "95"
> off of the GDAL package name, it sort of implies that this sort of mixing /
> matching would potentially be safe. I'm guessing that's not the case, but
> I figured it's worth asking.
Again, I did not hear any complaints from the field (yet), so... :)
Regards,
--
Devrim GÜNDÜZ
EnterpriseDB: http://www.enterprisedb.com
PostgreSQL Danışmanı/Consultant, Red Hat Certified Engineer
Twitter: @DevrimGunduz , @DevrimGunduzTR