Front page | perl.perl5.porters |
Postings from June 2006
On Tue, Jun 20, 2006 at 12:51:19AM +0200, Abigail wrote:
> OTOH, if we take minor releases into account, it situation even becomes
> worse. For instance, there were 20 minor releases in the two years between
> 5.005 and 5.6.0, but only 12 releases in the four years since 5.8.0. Between
> 5.004 and 5.005, there were 30 minor releases (not counting the more than
> a dozen trial releases), and Chip even had to invent a new counting system
> for the releases between 5.003 and 5.004.
I'm not disagreeing with your argument in general here, but I don't think that
the specifics this counting are valid, for a couple of reasons. I assume you're
referring to the chunk of releases here:
5.005 1998-Jul-22 Oneperl.
Sarathy 5.005_01 1998-Jul-27 The 5.005 maintenance track.
5.005_02-T1 1998-Aug-02
5.005_02-T2 1998-Aug-05
5.005_02 1998-Aug-08
Graham 5.005_03-MT1 1998-Nov-30
5.005_03-MT2 1999-Jan-04
5.005_03-MT3 1999-Jan-17
5.005_03-MT4 1999-Jan-26
5.005_03-MT5 1999-Jan-28
5.005_03-MT6 1999-Mar-05
5.005_03 1999-Mar-28
Leon 5.005_04-RC1 2004-Feb-05
5.005_04-RC2 2004-Feb-18
5.005_04 2004-Feb-23
Sarathy 5.005_50 1998-Jul-26 The 5.6 development track.
5.005_51 1998-Aug-10
5.005_52 1998-Sep-25
5.005_53 1998-Oct-31
5.005_54 1998-Nov-30
5.005_55 1999-Feb-16
5.005_56 1999-Mar-01
5.005_57 1999-May-25
5.005_58 1999-Jul-27
5.005_59 1999-Aug-02
5.005_60 1999-Aug-02
5.005_61 1999-Aug-20
5.005_62 1999-Oct-15
5.005_63 1999-Dec-09
5.5.640 2000-Feb-02
5.5.650 2000-Feb-08 beta1
5.5.660 2000-Feb-22 beta2
5.5.670 2000-Feb-29 beta3
5.6.0-RC1 2000-Mar-09 Release candidate 1.
5.6.0-RC2 2000-Mar-14 Release candidate 2.
5.6.0-RC3 2000-Mar-21 Release candidate 3.
5.6.0 2000-Mar-22
The two reasons I disagree are
1: The style of development has changed. It used to be that people patched
against the last numbered development release version. Hence there was a
need for fairly frequent development release versions. This was all before
the rise of rsync - these days everyone who pays attention gets their source
from rsync.
I've found there's no point in making snapshots because no-one even bothers
to download them. Release candidates don't fare much better, and I don't
think that we've had any feedback from the 5.9.3 release tarball itself -
everyone tracking blead does it via rsync.
2: You're counting development and stable releases there. Counting only stable
releases tells a different story. There were 3 maintenance releases of
5.005_0x before 5.6.0 was released. 5.005_01 was released within 7 days,
so for all practical purposes it *was* the stable release - treat 5.005 as
a de-facto release candidate. Heck. Even 5.005_02 was within 2 weeks.
So really there was 1 maintenance update, 5.005_03, about 6 months later,
1999-Mar-28. And then a MASSIVE 2 year gap before any further stable perl
release. And even then, for all Sarathy's hard work, 5.6.0 wasn't the
world's most stable Perl release.
So we had it worse 6 years ago. 2 years between *any* stable release.
And then only 1 stable release per year. (2001 - 5.6.1; 2002 - 5.8.0)
> New features is what the world uses as a measurement to decide software
> is dead, or being developed. And they look at what's being developed -
> for J. Random Perl Programmer, all the features currently in blead are
> just vaporware.
This I agree with. The world is hypocritical because it likes to know that
development is happening, but then it doesn't actually use the latest
version. But it's the world we live in.
Nicholas Clark