[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Release schedule

From: Tim Harrison
Subject: Re: Release schedule
Date: Wed, 26 Mar 2003 16:32:47 -0500
User-agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.2.1) Gecko/20021130

I think that the way I handle stable releases may be too complicated,
or perhaps it's that people are afraid to commit even bug fixes to the
release branch. So I am trying to think of different things to do the
next time. Just for your information, I typically make a pre-release
version and then fork a branch in CVS for bug fixes for the release. I
then make a release about a month after the pre-release.

From what I understand, a lot of people are concerned about sending in bugfixes that fix an immediate problem, but don't fix the underlying issues that lead to the immediate problem. Having a short term freeze only allows for people to spend a brief amount of time fixing, or at least, working around, obvious bugs. It might not provide enough time to deal with more subtle, or less than subtle, design issues.

One thing I might try next time is to make multiple (perhaps weekly)
pre-releases, so that people feel comfortable putting in bug fixes
knowing that they will be tested. Let me know if you have other ideas.

The more frequent the pre-releases, the less time there is to make major changes, if they're needed.

Most people update directly from CVS, without specifying a branch, from what I've seen. So, more than likely, the community at large isn't testing these branches. I don't have any numbers on this, I can only go by what I hear and see.

Also, after much procrastination, I hope to update the upcoming
goals/schedule for GNUstep:

If you have any thoughts on what our major goals should be for the next
few releases, let me know.

Might I suggest a more strict focus on making sure that the OpenStep aspects of the system are complete, clean, and fully functional? This way, at least the OpenStep functionality of GNUstep could be considered implemented. Even if it isn't perfect, at least the interfaces could be final, and would not change in the future.


Tim Harrison

reply via email to

[Prev in Thread] Current Thread [Next in Thread]