[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: postgis (2.5.2+dfsg-1) autopkgtest failure



Re: To Bas Couwenberg 2019-07-23 <[🔎] A6509D2A-027E-4D8A-9C6F-D7775A4DE2BE@df7cb.de>
> Am 23. Juli 2019 12:25:54 MESZ schrieb Bas Couwenberg <sebastic@xs4all.nl>:
> >But this will make pgapt unhappy, as they won't be able to co-install 
> >for other postgresql versions any more.

For other *postgis* versions. The packages for other postgresql
versions shouldn't be affected, independent from the postgis version.

> >Christoph, how bad would it be for pgapt if we change the Recommends to
> >Depends?

To recap, the problem with co-installing postgresql-X-postgis-Y-scripts
for different Y is that the packages all ship
/usr/share/postgresql/X/extension/postgis.control (and half a dozen
other .control files). The postgresql-X-postgis-Y (no -scripts)
packages do not have the problem because all files in there carry Y in
the filenames.

My plan is to move that file to postgis-Y.control, and install
"alternatives" so postgis.control will become a symlink that points to
the (by default) most recent postgis-Y.control file.

At the same time, dpkg diversions for .control are installed so an older
postgis version that hasn't been fixed yet can be installed in
parallel. (It will appear as postgis-old.control in the alternatives
tree.)

I did that for postgis-3, and if we think the idea makes sense, we
should also apply it to 2.5.

Christoph


Reply to: