[postgis-devel] PSC Vote: Drop the minor in the postgis lib files

[Regina Obe]

> Sandro Santilli <strk at kbt.io> writes:
> 
> > I guess you want to do this for bots to prevent them testing the wrong
> > library in a parallel tests run from different branches ?
[Regina Obe] 
No.  I want to do this only for speed regressions where I need to test both postgis versions with all else being the same except the version.
In those cases I usually have both installed in separate databases but same cluster and run a battery of tests on them and not at the same time.


> > But doing so would mean the bots won't test the default naming of
> > libraries. 
[Regina Obe] 
[Regina Obe] 
Only case would be Debbie and Winnie, but most  bots tests test from regress (not installed) so they wouldn't be testing the wrong library ever 
since the paths of the libs during regress are different
When extension is installed, even if it were possible that both tests are running for different branches (which I can make sure they don't),  
The create extension step would fail since we always add new functions to new releases (if it were testing against the older version).
Only case it would pass is if we are testing 3.0  and we are on 3.1

>> Maybe let's postpone this (and the switch needs a better
> > name anyway)
>
How about we don't postpone and you come up with a name you like or I'll go with the name I came up with and you can change later.
For most of the bots, there is no chance of them testing the wrong version even if I don't use this switch since they rebuild the cluster from scratch.

Which won't be the case for 1-2 years anyway.  

> Greg Troxel
> Regardless, tests need to be constructed so that installed libraries (those that
> are part of what's being tested) are not used, in general.
> It seems better to really fix this if it's needed, than to have a workaround.
[Regina Obe] 
Agreed we will be  testing both paths.


Thanks,
Regina