subversion-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Julian Foad <julianf...@btopenworld.com>
Subject Re: Aw: Status of ctypes-python bindings
Date Fri, 13 Feb 2015 12:30:58 GMT
Andreas Stieger wrote:
> Branko ─îibej wrote:
>> I've been looking at the ctypes-based Python bindings. They're barely
>> maintained; the only changes in the last two years seem to have been
>> minor tweaks to make the tests run. I'm also not aware of anyone using
>> them, or any packager bundling them.
>> 
>> This was an interesting experiment, but I think it's time to face the
>> fact that it failed.
>> 
>> If there are no objections, I propose to remove the ctypes-based
>> bindings from trunk (including all related configury and build system
>> support) before creating the 1.9 branch.
> 
> openSUSE packages them in 1.8 but I am not aware of (people) users of this 
> package, and I haven't seen other packages in the distribution using it. 
> Somewhat indifferent.

I'd be glad to get rid of it, on the basis that it's very far from complete and correct, and
doesn't appear to be much used, and that there are alternatives (mainly: swig-python, pysvn,
subvertpy).

However, there are probably people making good use of it. It's a bit harsh to delete a feature
completely, so close to release, with essentially no time for users to hear and react. If
I were a corporate user, with a team making use of it, and it disappeared with essentially
no notice, I would be sore. Deprecating it first and then deleting it at the beginning of
the next development cycle would be much better.

What is the benefit in deleting it from 1.9, compared with deleting it only from trunk? Is
there anything actively hurting us in keeping it around longer, other than the universal burden
that all extra code creates?

I would suggest we deprecate it, make it not built by the default "build all bindings" build
rules, and give people one minor release in which to migrate off it. Then delete it completely
soon *after* branching/releasing 1.9.

Perhaps we could contact the packager of the (one?) known distribution and see what they can
tell us? They might be glad to get rid of it, or they might know of particular user groups
relying on it.

- Julian


Mime
View raw message