Re: 3.4 CDs

[Miod Vallat <miod@online.fr>]

> Any "major" change has to be done within a couple of months, with a couple
> of months of testing and a couple of months of "freeze". Major changes that
> would take longer aren't possible.

This is complete bullshit. Any major change which would take more than
one release cycle to be stable enough to be worth using in production,
gets delayed. This does not mean that nobody will work on it.

For example, the i386 switch to ELF had been written long before 3.3 was
released, and could have been in 3.3. Except that there were also other
important changes, which would have made tracking regressions more
difficult.

And then, the advantage of a short release cycle like this is that, if a
functionnality misses a release, it's not a problem: the next release is
close enough. It's not a "now or in 5 years from now" kind of dilemna.

> There are too many times when development stops for releases.

No. Unless you do not consider intensive testing as not being part of
development.

> Too many human resources go into making releases. Or put another way, there
> aren't enough human resources to do releases properly.

The latter is true. And bitching about it will not help improve the
situation.

Miod