YAPC::EU::2008 in Copenhagen ended today. During the conference a lot of people said something like: Perl 5.10 is the best Perl release ever!. I think that who repeated this most was Damian Conway. Given his talks content I can understand why. But it wasn't the only one.
Now, what I can't understand is why http://ftp.cpan.org/src/README.html still says that perl 5.10.0 is a testing release, 8 months old.
I discussed this previously in perl 5 porters mailing list, but some months elapsed and no news. It is really important to re-tag that distribution with a stable tag, if it is really the best Perl release ever. Another option is to make 5.10.1 available. I know Rafael has some milestones for 5.10.1. But it might be more important for the Perl community to have 5.10.1 wide spread than to wait for really new cool features.
Why? For instance, because some Linux distributions will not include perl 5.10.x while it is tagged with testing. Also, because we all know that changes from 5.6 to 5.8 were basically from internals (Unicode, especially). But the changes from 5.8 to 5.10 are significally a breath of fresh air. We should make 5.10 the best Perl release ever for all users, and not just those geeks that want to use the new features.
I deeply respect and admire the work of everyone who's contributed to the Perl core, from tests to documentation to smoke results and bug reports. I still believe that the Perl 5 release process is badly broken, and the fact that the first "major" release of Perl 5 in five and a half years is a "testing" release is a symptom of that.
Re:Just My Opinion
sigzero on 2008-08-15T23:10:04
I can't use it at work because of that label.
Re:Just My Opinion
chromatic on 2008-08-15T23:24:19
Here's the pickle. People don't want to label their software as stable until other people test it and report it as stable. Other people don't want to run unstable software. As a consequence, alphas don't get tested. Betas don't get tested. Release candidates don't get tested. Stable testing releases don't get tested.
Release labels don't work. Long feature freezes which produce labeled releases don't work, at least if you want to produce stable software that people will use.
I owe the world a long rant about this, but I'll work on that later, and depress everyone who's been paying attention by saying "In addition, those other people don't upgrade, either."
Re:Just My Opinion
sigzero on 2008-08-16T16:35:26
Oh I agree. "I" would definitely use it at work. I keep hammering away at doggedly stubborn management.
Re:Just My Opinion
Alias on 2008-08-16T22:26:22
> Stable testing releases don't get tested.
90% of Strawberry Perl downloads are the stable testing release (500-1000 a day)
10% of Strawberry Perl downloads are the stable release (50-100 a day)
Dunno about the rest of the platforms, but the Windows world is testing the hell out of 5.10.0.
Re:Just My Opinion
chromatic on 2008-08-17T05:11:15
That's really interesting! Do you have a conjecture as to why this is? I'd love to know how to duplicate this for other projects.
Re:Just My Opinion
gabor on 2008-08-17T09:04:44
You did not ask me but here is my take on it:For one, the Strawberry web site does not say testing and all things equal people tend to go with the higher version.
IMHO In the Windows world it is much more common to download your software while most of the linux/UNIX people will go with the one that is in the distro.
Besides we don't have numbers for download to Linux/Unix.Re:Just My Opinion
chromatic on 2008-08-17T16:32:46
For one, the Strawberry web site does not say testing and all things equal people tend to go with the higher version.
Ah, there's the biggest reason.
