Committer nominations - Indra Talip and Stefan Richthofer

I would like to nominate both Indra Talip and Stefan Richthofer to become
committers on Jython.

I have personally worked with both Indra and Stefan in getting their
patches and pull requests accepted and in trunk. In both cases, they have
made fantastic contributions, presented their work at Python conferences,
been involved in Jython sprints, submitted bugs, and been active on our
mailing lists. They are clearly qualified to join our group of committers,
and I hope to see this level of contribution to continue with us in this
new role.

*Indra Talip* has made numerous contributions to Jython, including on
fixing ClassLoader leaks, memoization of codepoint computations in regular
expressions, and in fixing defaultdict. In particular, I had a great time
collaborating with Indra on defaultdict and making it simpler, which also
allowed us to upgrade our inclusion of Google Guava collections to the
latest version.

I asked Indra to share something about his favorite bit of work on Jython:

The bz2 fixes came about because I was interested in contributing to Jython
> following my participation in the sprints at pycon.au in 2013 where I
> contributed to CPython but felt that my skills were better suited towards
> Jython. Having started using Jython by embedding first the 2.7alpha1
> release and then a custom build from trunk, in order to pick up fixes for
> memory leaks, codecs having static state etc, I felt that it would be great
> if Jython could reach a full 2.7 release. As such I thought that the best
> way of making that happen faster and to learn how Jython worked under the
> hood was to attempt to fix some unit tests. The bz2 module seemed simple
> enough to understand however I discovered that the rabbit hole was a little
> deeper than I anticipated and before long I was asking my friends to talk
> me out of refactoring it to use Jython's PEP3116 infrastructure which I
> saw as a) simpler to understand and b) where/how the bz2 module was
> implemented Python3 . Their response was to "just pack a parachute and
> jump". The satisfaction of seen that land and fix so many failing unit
> tests in not only the bz2 module but also the tar module was great :)


Landing bz2 was a significant milestone for Jython. We would not be able to
fully support pip, or wheel format, without this work by Indra.

Bio. Indra is a senior software developer with twelve years of experience
with the Australian government. On any given day, he develops or maintains
software using Python, Java, C, C++, Javascript or PHP. Notable systems
that Indra has worked on, supported and developed include a grid computing
distributed processing system, full-text search application and most
recently embedding Jython into a bespoke Java framework.

*Stefan Richthofer* is the author of the JyNI project, which seeks to add
complete C extension API support for Jython. But this has also meant that
Stefan has also started to look at what Jython needs to (minimally) do to
support JyNI in terms of additional hooks. As a consequence, Stefan and I
have collaborated on some of the most challenging aspects of the Jython
runtime. I particularly enjoyed the chance to sprint with him in Aachen
Germany in August where we found ourselves discussing
http://bugs.jython.org/issue1057 ("new style classes don't support
__del__") in the context of these hooks. In his own words:

we wondered whether this issue could be solved with this approach too. In
> the following days I worked out a fix for issue 1057 in a way that
> fulfilled JyNI's hooking requirement at the same time, taking care that the
> new finalization process would not interfere with JyNI's planned garbage
> collection approach. The finalizability of new style classes is now
> handeled by such finalize triggers. This way, expensive finalizable objects
> (the triggers) are only created for instances that really need them.


Landing this support for __del__ means that objects that expect finalizer
support for resource cleanup now actually work on Jython. This is a huge
thing for us, especially for what we are doing in 2.7 with resource leaks
as a whole. Stefan's ability to execute with strong testing this
significant change indicates he is ready to be a committer.

Bio. Stefan graduated 2009 at Ruhr-University Bochum/Germany in Mathematics
(minor: Computer Science). Since then he has done research in computational
neuroscience at the Institute For Neural Computation, suited also at
Ruhr-University. For his research on Predictable Feature Analysis Stefan
uses Python extensively, especially NumPy and MDP (Modular Data Processing
Toolkit). Still he considers Java to be his native programming language and
did some projects with it. That is why he is excited about Jython and
Java/Python integration in general. In May 2013 he started the JyNI project
which aims to enable Jython to use native C-Python extensions (www.jyni.org
).

- Jim

------------------------------------------------------------------------------
Want excitement?
Manually upgrade your production database.
When you want reliability, choose Perforce
Perforce version control. Predictably reliable.
http://pubads.g.doubleclick.net/gampad/clk?id=157508191&iu=/4140/ostg.clktrk

_______________________________________________
Jython-dev mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/jython-dev