Differences between revisions 1 and 2
⇤ ← Revision 1 as of 2007-01-07 00:31:53 →
Size: 109
Editor: EricDobbs
Comment:
← Revision 2 as of 2007-01-07 01:29:58 → ⇥
Size: 2738
Editor: EricDobbs
Comment:
Deletions are marked like this. Additions are marked like this.
Line 1: Line 1:
== Getting to a 2.2 release ==

Charlie: I think '''slots''' are the last feature we need, so once that lands we can
release a beta. There are a few remaining test failures in
regrtest.py that I think we need to get before we can make a real
release. Those two things and the inevitable bugs that come up from
getting the code out in public are the only things between us and a
release.

Samuele: one thing I think is still missing is '''proper `__del__` support for user-defined new style classes'''.

(Extracted from [http://sourceforge.net/mailarchive/forum.php?thread_id=30988615&forum_id=5587 Notes as I bootstrap into Jython])


== What's needed for 2.3 ==

Samuele: The next big things that we can foresee are indeed '''finding a solution for jythonc''' and
'''improving our overloaded java methods resolution''' so that it can behave properly once we introduce bools.

(extracted from [http://sourceforge.net/mailarchive/forum.php?thread_id=30988616&forum_id=5587 Re: Notes as I bootstrap into Jython])


== Componentizing JAR Indexing ==
Line 2: Line 26:

Charles asked: How much effort would it take to '''componentize the whole jar-indexing'''
subsystem? We'd be interested in using it for "import" purposes in JRuby

Charlie replied: the indexing code could use a significant cleanup. It's mostly contained
in the PackageManager class hierarchy. SysPackageManager extends PathPackageManager which extends
CachedJarsPackageManager which extends PackageManager. Disentangling the interactions between that mess of inheritance makes working on the
caching system harder than it should be.

Charles added: '''making the dynlangs be able to call across each other''' without some awful marshalling layer to/from java. In other
words, we should be able to pass an object from JRuby to Jython to Rhino and read fields and invoke methods without difficulty. That's where a
common underlying runtime or set of interfaces will be an absolute necessity.


== Compelling Goals ==

(extracted from [http://sourceforge.net/mailarchive/forum.php?thread_id=30988621&forum_id=5587 Setting a compelling goal]) Charles: Just supporting Python is not enough of a goal to compel folks to contribute precious
off-hours to the project. Working toward support for apps like Django
would do a lot more for publicity and project interest.

 * django
 * TurboGears
   * SQLObject and/or SQLAlchemy
   * CherryPy
   * Kid and/or Genshi
 * Twisted (using [http://www.xhaus.com/alan/python/jynio/select.html Alan Kennedy's select implementation])
 * WSGI (using [http://www.xhaus.com/modjy/index.html modjy])

Getting to a 2.2 release

Charlie: I think slots are the last feature we need, so once that lands we can release a beta. There are a few remaining test failures in regrtest.py that I think we need to get before we can make a real release. Those two things and the inevitable bugs that come up from getting the code out in public are the only things between us and a release.

Samuele: one thing I think is still missing is proper __del__ support for user-defined new style classes.

(Extracted from [http://sourceforge.net/mailarchive/forum.php?thread_id=30988615&forum_id=5587 Notes as I bootstrap into Jython])

What's needed for 2.3

Samuele: The next big things that we can foresee are indeed finding a solution for jythonc and improving our overloaded java methods resolution so that it can behave properly once we introduce bools.

(extracted from [http://sourceforge.net/mailarchive/forum.php?thread_id=30988616&forum_id=5587 Re: Notes as I bootstrap into Jython])

Componentizing JAR Indexing

[http://sourceforge.net/mailarchive/forum.php?thread_id=30988620&forum_id=5587 Componentizing JAR Indexing]

Charles asked: How much effort would it take to componentize the whole jar-indexing subsystem? We'd be interested in using it for "import" purposes in JRuby

Charlie replied: the indexing code could use a significant cleanup. It's mostly contained in the PackageManager class hierarchy. SysPackageManager extends PathPackageManager which extends CachedJarsPackageManager which extends PackageManager. Disentangling the interactions between that mess of inheritance makes working on the caching system harder than it should be.

Charles added: making the dynlangs be able to call across each other without some awful marshalling layer to/from java. In other words, we should be able to pass an object from JRuby to Jython to Rhino and read fields and invoke methods without difficulty. That's where a common underlying runtime or set of interfaces will be an absolute necessity.

Compelling Goals

(extracted from [http://sourceforge.net/mailarchive/forum.php?thread_id=30988621&forum_id=5587 Setting a compelling goal]) Charles: Just supporting Python is not enough of a goal to compel folks to contribute precious off-hours to the project. Working toward support for apps like Django would do a lot more for publicity and project interest.

JythonDeveloperGuide/PleaseAdoptMe (last edited 2008-11-15 09:15:58 by localhost)