That "vision" thang...
Stephen J. Turnbull
stephen at xemacs.org
Tue Mar 3 20:45:17 EST 2009
Mats Lidell writes:
> >>>>> Stephen wrote:
> Well it would make sense to me to go GPLv3 first if that would make it
> easier to do a release.
It doesn't make it easier to do a release. It makes it easier to
convert packages and to sync. Remember, some developers are not going
to care about the release, but would possibly be willing to contribute
syncs of packages they care about a lot.
> Stephen> GPLv3 and GNU sync are not necessary to a release. If people
> Stephen> work on GPLv3 in parallel with the technical problems, we can
> Stephen> release as GPLv3. If not, we don't have to.
> Yes, but as a project I think we need a plan
As a project we need a release manager!
A plan that nobody intends to follow is useless. We have a number of
separate tasks, at least the following:
- fix really ugly bugs in 21.5
- make some attractive features in 21.5 properly customizable
- convert the license of 21.5 to GPLv3
- convert licenses of packages that want to do so to GPLv3
- sync internals (eg, jit-locking) to recent GNU
- update packages that are stalled because they depend on recent GNU
- fix more bugs in 21.5 (eg the intermittent memory leaks)
- finish up projects (KKCC, the incremental GC, window configurations)
- add features that are "almost ready now" (curl and neon bindings,
libffi are in my workspaces; I imagine others have their own
projects; many SXEmacs features could be ported "cheaply")
that are technically independent. We need to decide which are
required for a release, and then we can start planning the release.
People who want to work on technically independent tasks rather than
the release should be allowed to do so, but they should also be made
aware that they're not helping with the release.
> Stephen> General GNU Emacs sync is an ideal I aim at, not a concrete goal I
> Stephen> expect to ever achieve. That can go on in parallel with other
> Stephen> development as soon as we have a GPLv3 tree.
> Yes, but at times we could make an effort and focus on some areas
> where syncing would buy us much. Is this such a time?
I don't know. You seem to think a quick release is important. But
technically a "sync" is not a prerequisite to a release, and putting
it into the list of release tasks will delay release by many months.
In fact the only absolute requirement for a release is a release
manager. I can do it, but I doubt you'll see a public release this
year if I do, unless people start contributing code in a hurry. I'm
not in a hurry to release, myself, but I would support others if they
do the work they think is needed for a release.
More information about the XEmacs-Beta