Re: Release date of Emacs 24.4?

[Óscar Fuentes]

Drew Adams <drew.adams@oracle.com> writes:

>> AFAIK a release is planned to happen in a few weeks.
>
> Really?

Not sure. The maintainer's explicit goal is to speedup the release
cycle. Under this maintainer, past release's pretest phase was much
shorter than under RMS. I wouldn't be surprised if the release is made
two months after the end of the feature freeze.

> There hasn't even been a pretest announced and distributed,
> has there?  Where is the announcement asking people to test the
> pretest distribution?
>
> What about the pretest testing period?  Has that been reduced to
> at most a few weeks now?
>
> There are thousands of bugs reported now, including many newly
> introduced.  Do we really intend to ship a release only a month
> or two after closing the addition of new features?

That was not what I was suggesting. Emacs is on feature freeze for 6
weeks now (IIRC) and it seems to me that the maintainer wish to
stabilize (*) it enough to turn the pretest phase into a mostly
packaging test exercise.

I think this is a sensible approach. How much pretesters do we have,
compared against the number of users who use the development version of
Emacs on a production environment? Packaging and exotic platforms aside,
the pretest phase discovers a tiny amount of bugs.

> Not adding any more new features is one thing.  Actually testing
> those new features, and regression-testing old features, and
> fixing known bugs are something else again.

If you expect from the pretesters a thorough testing of Emacs, you need
a *very* long pretest phase (á la RMS). It is much more effective to
release the beast on a "good enough" (**) state and wait for the bug
reports from the multitude of users. Then fix those bugs and do a point
release. Conservative users can wait for the dot release, which I'll bet
that it will happen sooner than the "long pretest" release. And will be
more stable too.


(*) After using the development version of Emacs for 10+ years, I can
say that it is one of the most reliable software packages I ever
installed on any of my machines.

(**) No critical bugs, no significant regressions.