CPANdeps can now filter results by OS. Because people asked for it. Personally, I don't think it's very useful. The purpose of the site was originally to help module authors determine which modules are good to depend on because they're easy for your users to install. It's rare for a module author to have control over what OS his users are suffering from (the obvious exception being platform-specific modules like Solaris::Procfs, BSD::Resource, or Win32::DriveInfo) and so filtering by OS isn't helpful.
Not sure how useful it is, but it is interesting to see the differences between OSs. Data::Phrasebook has a 50% chance of success on Win32, a 76% chance on Linux and 100% OpenBSD. (It would be 100% across the board if Test::Harness could drop File::Spec, but that's another story).
BTW
Re:Interesting though
drhyde on 2007-11-29T15:27:19
Oi! There's nothing wrong with glam! Not even Welsh glam! It's on the list cos the tape I bought years ago has broken.
If you want to see where code fails, then I think Slaven Rezic's matrix thingummy is a better tool. If only I could remember the URL!
Re:Interesting though
srezic on 2007-11-29T20:28:18
Here it is. Maybe I should ask for a link from the official cpantesters site? Or just contribute my cgi?Re:Interesting though
barbie on 2007-11-30T08:15:18
That looks great. However, as a feature request could you also include 5.6.1. The reason being that ActiveState ship 5.6.1 not 5.6.2, so would have more meaning to those who are not able to upgrade perl on Win32 boxes, but can upgrade module distributions.
Re:Interesting though
srezic on 2007-11-30T22:02:01
The script shows all perl versions with CPAN tester reports for that module. It seems that there are almost no testers around with a 5.6.1 setup... correct, only about 30 test reports in the last month.Re:Interesting though
barbie on 2007-12-02T00:42:50
All my 7,000+ test reports were for 5.6.1, Corion and others have tested on 5.6.1 too.Re:Interesting though
drhyde on 2007-12-03T21:02:30
<testresults> select perl, count(perl) from cpanstats group by perl;
perl,count(perl)
'5.10.0',39700
'5.3',2
'5.4',1020
'5.5',7703
'5.6.0',9926
'5.6.1',32952
'5.6.2',27536
'5.7.1',496
'5.7.2',23
'5.7.3',236
'5.8.0',41683
'5.8.1',16059
'5.8.2',11030
'5.8.3',20045
'5.8.4',24223
'5.8.5',62332
'5.8.6',45105
'5.8.7',71819
'5.8.8',249892
'5.9.0',468
'5.9.1',2208
'5.9.2',1804
'5.9.3',2667
'5.9.4',5402
'5.9.5',63927
'5.9.6',33
[26 rows of 2 fields returned]
5.10.0 isn't even out yet and has more results than 5.6.1:-) Re:Interesting though
barbie on 2007-12-04T10:21:26
Only because you, Chris, Andreas and Slaven have choosen to ignore it from your test environments
;) Although to be fair Win32/5.6.1 is probably the one area of testing that is really needed, as it still has a major automated installer (ActivePerl 5.6.1 build 638) available, and probably still has a significant user base on that platform. The fact that 5.10 has more tests only highlights how much test data we are lacking on 5.6.1.
Re:Interesting though
drhyde on 2008-07-09T23:49:11
I thought it might be interesting to see the same query's results seven months later...
<testresults> select perl, count(perl) from cpanstats group by perlperl,count(perl)
'5.10.0',450893
'5.11.0',152126
'5.3',2
'5.4',1020
'5.5',18068
'5.6.0',9926
'5.6.1',33123
'5.6.2',83297
'5.7.1',496
'5.7.2',69
'5.7.3',267
'5.8.0',42371
'5.8.1',31796
'5.8.2',22576
'5.8.3',32110
'5.8.4',33703
'5.8.5',79174
'5.8.6',66565
'5.8.7',90809
'5.8.8',541146
'5.9.0',1157
'5.9.1',2215
'5.9.2',2236
'5.9.3',3854
'5.9.4',5731
'5.9.5',66673
'5.9.6',33
[27 rows of 2 fields returned]Re:Interesting though
drhyde on 2007-11-30T15:17:51
Thanks, I'll add links to your site for each module in the dependency trees.
