[IronPython] Ironpython 0.7 released!
jonathan-lists at cleverdevil.org
Tue Mar 29 00:34:00 CEST 2005
Jim, and others,
> I'm in complete agreement that we've got work to do on the IronPython
> community site. This feedback is great.
I think that "work to do" is probably an understatement. I think the Passport
thing is probably a showstopper for many Open Source people. I am not going to
argue the technical merits for this, but to many OSS people, Passport is
considered some great evil. It doesn't really matter if there is a technical
reason or not, that is the way it is perceived. It doesn't matter how good the
site is, its not going to get used if Passport is required.
This is obviously a problem. There are other, better, alternatives out there,
although I am sure that you don't have a huge choice in this issue being that
gotdotnet is Microsoft's baby. I can't blame you here, but its a hurdle to be
sure. Maybe you could make it possible for people to file bug reports with
just a valid email address, and no passport? This might solve the problem.
> I've had my first experiences using the forums this last week, and I'm
> not particularly excited about them either. One thing that we're
> investigating is NNTP access to the gotdotnet forums. If we can get
> this setup I hope that people will give that a try and see how that
> works for them. We're going to keep looking at those kinds of options
> to make the gotdotnet site work better for people. However, so long as
> you find this mailing list to be the best way to discuss IronPython then
> it won't be going away.
The forums aren't good, and probably never will be as good as a mailing list.
NNTP is probably worse, because while it is more like a mailing list, its
harder to access. You said it yourself: a mailing list is the best way to go.
SourceForge projects have both available, and 90% of projects (in my estimation)
use their mailing lists a lot, and their forums a little. I would say to shut
down the forums, and make the mailing list the appropriate way to deal with
This will please your target community greatly. I for one can't think of
anything better than a mailing list here.
> I agree that the bug database could also use some improvement, but it
> would be great if we could figure out how to make the current version
> work for people today. Some people will see needing a .NET passport as
> a hurdle to filing bug reports and any hurdles we add here are a
> mistake. Nevertheless, have you tried to get a .NET passport? It's a
> fairly simple process and it should work on any OS. While this is a
> little cumbersome, I don't think that it's much different from most bug
> trackers that will require a valid email address before accepting bug
> reports. Do you have a technical reason that you don't want to/can't
> get a passport to file bugs?
What I said above still applies: make it work without passport, and we will be
happy. Otherwise: you will lose a very large portion of your community. It
doesn't matter if there are good reasons (technical or political), all that
matters is the result. I want IronPython to have a strong community, and I am
sure that Microsoft and You do too!
The issue that you haven't addressed in this email is the license. I know its
probably because you have some serious discussion to do, but I want to go ahead
and register my agreement that the license has some issues that would be really
easy to fix. I agree with you that the license itself is fairly
straightforward, but you would be much better served by picking ANY OSI
approved license. As far as I am concerned, your target audience with
IronPython is the Python community and the Open Source .NET community. By and
large, this means Python people and Mono people. I am in both camps, and I
would have to say that I would be totally satisfied if IronPython used a
simple, non-viral BSD like license. If you adopted the same license as the
Mono folks use, I think it would go really really far, and you would
immediately cause all of the issues to disappear.
Just to show you the potentially disastrous effects of the choice, take a look
at what is already happening:
Edd Dumbill is a Mono hacker who really liked IronPython. Now, he evaluating
Boo, another dynamic .NET scripting language, instead. This guy wrote a .NET
book, and might even write a book about IronPython (which would be a big deal
if you ask me) if he was able to get over the license thing.
I am really worried that you are missing an opportunity here. We all have great
respect with what you have done so far, but I bet you that most of us would be
fairly unlikely to contribute given the current state of affairs. If you ask
me, a simple license change, keeping the mailing lists, and not requiring a
passport to file bugs would solve most of the issues.
The other issue is your statement about external people not being able to
contribute patches. This is a big problem. I won't file bugs if you don't
give me the opportunity to help out in a tangible way. Its clearly not an open
source project then: its "shared source" to the core. Which do you want to be?
Decide so that I can choose to participate or not. Personally, I don't see why
you can't have a public SVN or CVS repository containing IronPython. Accept
patches, review them (you don't have to apply bad ones). If you don't have
time to review them, find at least one more person you trust to be a committer.
If you really wanted to impress: make that additional committer a
well-respected mono hacker. This would truly show the community that you guys
are committed to making IronPython the best .NET scripting language on any
So, to review my pie-in-the-sky wishes to guarantee IronPython success:
1. Switch to a similar, but respected license.
- Bonus points for picking the same license as mono.
2. Ditch the gotdotnet forums. Make the mailing list "the way".
3. Don't require a passport to file bugs.
4. Have a public SVN or CVS repository.
5. Be open to accepting external patches.
6. Have at least one other "committer" (to help lighten the load on you).
- Bonus points if this person is outside Microsoft.
- Extra bonus points if this person is a respected Mono hacker.
Items 1, 3, and 5 are non-negotiable to me. If you don't do these at least, I
guarantee that the project will stagnate, and something like Boo will become
more widely accepted (especially by Open Source types).
Anyway. I don't want to criticize you too much, as I think IronPython is a great
piece of work. I would just hate to see such a great opportunity be missed!
More information about the Ironpython-users