[geos-devel] Petition for a 2.2.2 release

[Sean Gillies]

On Mar 24, 2006, at 2:52 AM, strk at refractions.net wrote:

> On Thu, Mar 23, 2006 at 08:34:47PM -0700, Sean Gillies wrote:
>
>>> Please take a look at the HOW_TO_RELEASE file in HEAD.
>>> It contains information on the process. It's a pretty
>>> new file, so let me know if you have any question you'd
>>> like to discuss.
>>>
>>> Note that it has been written w/out thinking about a 2.2.2 release,
>>> so I'm not sure about implications of the 2.2.2 release.
>>>
>>> In particular, in HEAD we decided to name C++ library after release,
>>> with every release breaking ABI with previous (libgeos-2.2.2.so).
>>>
>>> The 2.2. branch shouldn't have this big change, so we should make it
>>> binary-compatible with the previous, thus only incrementing MICRO
>>> version (libgeos.so.2.2.2). We should check if everything goes as
>>> expected still followin the same release process.
>>>
>>> --strk;
>>
>> Alright, I've read it and seems fine, but your uncertainty about the
>> implications confuses me. I've incremented VERSION_PATCH in
>> configure.in to 2. Make install results in libgeos.so.2.2.2 and
>> libgeos_c.so.1.0.2. I've made dist and verfied that the tarball will
>> build and install. Can we tag it and release?
>
> Sounds good to me, as a final test can you run 'make dist', build
> and install from the resulting package ? Keep an eye to swig.
>
> --strk;

Done. Checks out. Swig Python module builds as well, but I'm starting  
to think that the swig interface should be discouraged for the same  
reasons you discourage use of the C++ API. There's no stability to it.

Sean
---
Sean Gillies
http://zcologia.com