[261] in BarnOwl Developers
Re: Regarding owl development
daemon@ATHENA.MIT.EDU (Alejandro R. Sedeno)
Thu Oct 29 18:04:15 2009
Resent-From: nelhage@mit.edu
Resent-To: barnowl-dev-mtg@charon.mit.edu
Date: Wed, 01 Nov 2006 16:06:49 -0500
From: "Alejandro R. Sedeno" <asedeno@MIT.EDU>
Reply-To: dirty-owl-hackers@MIT.EDU
To: "James M. Kretchmar" <kretch@mit.edu>, Erik L Nygren <nygren@mit.edu>
CC: dirty-owl-hackers@mit.edu
In-Reply-To: <20061031061328.GA16462@mit.edu>
I'm curious about some of the changes you've listed, in particular the
rewrite in python. Do you mean replacing the perl engine with a python
one, or replacing the C code with python (presumably taking out perl in
the process)?
The general feeling I get from our (d-o-h) list and zephyr traffic is that
collaboration seems more and more unlikely, but I'd like to at least have
ideas flow between projects, keeping some of the basic functionality
compatible.
Ways that we've hooked Perl into C:
* Hook owl's main loop into Perl by way of owl::mainloop_hook()
* Added a mechanism by which Perl can register new commands complete with
description and summary information.
* Refactored the editwin code, allowing for the text to be passed VIA a
callback.
* Added owl::start_edit_win() to allow Perl implementation of commands
like jwrite for the Perl-based jabber module. This uses the callbacks above.
* Added a constructor to owl::Message, and the method owl::queue_message()
to allow for message injection from Perl.
* Added a hook into the buddy list code to call out to owl::get_blist()
for externally implemented protocol buddy-lists.
Probably some more that I'm forgetting. (d-o-h: please feel free to add to
this)
We're also working on polishing the modular owlconf system I developed to
use actual Perl modules, pulling much of its core functionality into
perlwrap.pm, and setting up additional Perl include paths for owl-specific
libraries for the installation and the user.
I'm also planning on adding calls from the style code to Perl-implemented
protocols, allowing styles to ask for a default message format. This will
allow the default style in C to format an jabber message without the need
to tweak what gets placed into the standard message attributes for it to
look good, and without giving C knowledge of protocols implemented in
perl. I've also got analogous ideas for logging, and other tasks might
want similar behavior.
A long term idea that we may or may not get to is pulling all
protocol-specific code out of owl entirely - reimplementing AIM and Zephyr
in Perl - allowing the C side of owl to become a very clean and robust
generic Instant Messaging UI.
If we do end up forking, our plans are to call the project BarnOwl,
because, well, we'd like to keep owl with all our ponies.
-Alejandro
James M. Kretchmar wrote:
> Ok, so a few topics here:
>
> 1. Future plans for owl
> 2. Collaborative development
> 3. Specific changes so far
>
> First future plans, as you asked for. Future plans for Owl look
> something like:
>
> - Outstanding bug fixes
> - OTR support
> - Jabber support
> - (possible) Generic gaim plugin support
> - Replace AIM library
> - AIM chatroom support
> - Other protocol support (IRC etc.)
> - Rewrite in python
> - Rearchitect to separate UI from message engine
> - Curses UI improvements (list selectors etc.)
>
> although it's not fixed in stone. Also some of those are
> interdependent. Generic gaim plugin support might obviate the need
> for replacing the AIM library, and would definitely eliminate the need
> for direct support of other protocols like IRC. A rewrite in python
> would include the work to better separate the UI from the message
> engine.
>
> A few things have been in the works, although as you rightly point
> out, very slow to progress. I was gearing up a release for
> outstanding bug fixes and OTR support a couple of months ago, but the
> whole thing stalled out when the OTR support turned out to be buggy.
> I also heard from a strong developer that he might have official time
> to add jabber support to Owl in the near future, although I don't know
> if that will pan out or not.
>
> Second topic: collaborative development. I've definitely been the
> bottleneck on development for a while. Ideally, I'd like to keep a
> single owl project line since diverging projects often cause pain for
> the common user and it's been an explicit philosophy of owl from the
> start to function well for both the expert and the novice. I can see
> two paths from here:
>
> 1. If we find we're definitively or irreconcilably at odds over the
> architectural direction of owl there won't be any choice but to
> fork off a new project. My guess is this isn't the case, but I
> don't know for certain. If it is the case we should talk about
> how to work out naming your branch to avoid future confusion.
>
> 2. If we find that we're all more or less on the same page with the
> architectural directions I want for Owl, then I think we mostly
> have a problem with the mechanics of making collarborative
> development work. I'm open to all suggestions here as I'm
> clearly getting too used to the corporate world. Maybe we should
> open up much wider access to write to the CVS respository. If we
> do that along with discussion in CRs about proposed
> implementations before coding, and have good pre-beta testing
> that might work out fine.
>
> Third topic: specific changes you've got. Tell me more about the
> changes you've made. Exposing more of the C functionality to the perl
> config sounds fine as long as it's not something that's going to
> obviously break configs in future versions when we change C
> functionality. What kind of things are exposed? Also, using that to
> work in Jabber support through Perl is very nifty, although I don't
> think we should consider that "real" jabber support (i.e. if folks
> want to use that to define a jwrite in their owlconf that's cool, but
> I wouldn't want to distribute owl with a built-in jwrite that relies
> on Net::Jabber).
>
> BTW, I've cc'd nygren who's been heavily involved with parts of owl
> development in the past.
>
> kretch
>
> On Mon, Oct 30, 2006 at 01:08:04AM -0500, Alejandro R. Sedeno wrote:
>> kretch,
>>
>> The owl hacker community appears to have achieved some critical mass
>> last week. We've taken owl from cvs and dropped it into an svn
>> repository where have been hacking on it together. We've incorporated
>> the patches on bugzilla, and extended its functionality in new and
>> interesting ways.
>>
>> We would like to collaborate on owl, but from our point of view owl
>> development appears to have stagnated; the last release is already over
>> a year and a half old. We would like to know what your plans are for owl
>> before we fork and release our own version.
>>
>> We'd also like to share some of the ideas we've been working on. Over
>> the course of the last week, we've exposed more of owl's C functionality
>> to perl, allowing it to hook into main loop and inject arbitrary
>> messages. We've taken that base and and the owlconf system I (asedeno)
>> mailed you about in Jan 2005, and in the past couple of days have added
>> preliminary jabber support using Net::Jabber.
>>
>> The svn repository is available at
>> /afs/athena.mit.edu/user/n/e/nelhage/svn/dirty-owl-hacks/
>> readable by the public LIST:dirty-owl-hackers.
>>
>> We look forward to hearing from you,
>>
>> Those dirty owl hackers,
>> dirty-owl-hackers@mit.edu
>>
>
