[postgis-devel] liblwgeom.so

[Mark Cave-Ayland]

On 09/09/11 08:58, Sandro Santilli wrote:

>> Because we know they work, because the number of potential
>> install-time mistakes that can break things goes down, because the
>> build chain gets simpler. Why would we not?
>
> I know the "we know it works" argument very well, but you have to draw
> a line between stability and innovation. Given we're going 2.0 (how many
> things known to work we left already?) it seems a good occasion to move
> forward to me.
>
> About mixing static linking and dynamic linking, I left the postgresql
> module statically linked to avoid the complexity of pre-installation
> testing which would require (for a dynamically linked module) a way to
> have the already running PostgreSQL instance
> find a yet-to-be-installed library. I'm not sure that's possible at
> all. In order to fix that we should rework the tester to at least fire
> a completely new instance of postgres (which could actually also fix
> other inconsistencies, like the need to be superuser there).
> Beside regress testing, a dynamically linked module would also require
> postgresql restart after installation (and after the ldconfig call).
> I don't see how mixing dynamic and static linking is a problem at the
> moment. After all the dynamic libary is versioned after PostGIS and
> tightly coupled with the specific release. It's like a convenience
> internal library for now.

I think all these problems are solvable, however now we currently have 
the worst of both worlds which is that we have to support 2 separate 
ways of building everything - we need to stick with one method for doing 
this, and my choice would be dynamic.

> About taking responsibility for the libpgcommon issue I agree it is on
> me. I'm just back from holiday and will be on it. Mark: I asked for
> your opinion because I belive it has to do with using or not PGXS, but
> I could be wrong. Eventually I'd drop libpgcommon and simply link its
> object files into the respective libraries. It's been a mistake on my
> side to commit the libpgcommon thing while working on liblwgeom,
> raising the warning levels. Sorry for that.

Indeed. I still maintain that the libpgcommon patch should not have been 
applied in its current form, and should be removed before final 2.0 release.


ATB,

Mark.

-- 
Mark Cave-Ayland - Senior Technical Architect
PostgreSQL - PostGIS
Sirius Corporation plc - control through freedom
http://www.siriusit.co.uk
t: +44 870 608 0063

Sirius Labs: http://www.siriusit.co.uk/labs