[gentoo-dev] Package ranking by number of ebuilds on the portage tree
So I have been doing some bash scripting out of some comment in a conversation to count (and rank) the packages by the number of ebuilds they have (and thus of versions of said package). The results can be seen at http://dev.gentoo.org/~klondike/ebuildrank.txt and if there is interest I can try to automate the generation of the ranking daily (though I'd like infra's comments on that). As some extra candy I have calculated which is the percentage of packages holding the same number of packages, it is impressive seeing how more than 50% of the ebuilds have only 1 and more than 75% have two or less! If I recall my stats classes the number of versions in a package seems to follow some kind of poison distribution. 175 .00635687496026953100 68 .00635687496026953100 31 .00635687496026953100 28 .00635687496026953100 25 .00635687496026953100 23 .01271374992053906200 22 .01271374992053906200 21 .01271374992053906200 20 .01271374992053906200 19 .01271374992053906200 16 .05721187464242578300 15 .04449812472188672000 14 .04449812472188672000 13 .07628249952323437700 12 .08263937448350390900 11 .12078062424512109800 10 .15892187400673828700 9 .22884749856970313300 8 .31784374801347657400 7 .76282499523234377900 6 2.20583561121352742900 5 2.47282435954484775200 4 4.54516559659271502100 3 11.39151992880300044400 2 25.75170046405187209900 1 51.64325217722967389200 Hope you enjoy the numbers as much as I enjoyed calculating them! klondike signature.asc Description: OpenPGP digital signature
Re: [gentoo-dev] Package ranking by number of ebuilds on the portage tree
On Fri, Oct 26, 2012 at 2:02 PM, Francisco Blas Izquierdo Riera (klondike) klond...@gentoo.org wrote: So I have been doing some bash scripting out of some comment in a conversation to count (and rank) the packages by the number of ebuilds they have (and thus of versions of said package). The results can be seen at http://dev.gentoo.org/~klondike/ebuildrank.txt and if there is interest I can try to automate the generation of the ranking daily (though I'd like infra's comments on that). We can put the script in qa-reports.g.o
Re: [gentoo-dev] Package ranking by number of ebuilds on the portage tree
On Fri, Oct 26, 2012 at 1:16 PM, Theo Chatzimichos tampak...@gentoo.org wrote: On Fri, Oct 26, 2012 at 2:02 PM, Francisco Blas Izquierdo Riera (klondike) klond...@gentoo.org wrote: So I have been doing some bash scripting out of some comment in a conversation to count (and rank) the packages by the number of ebuilds they have (and thus of versions of said package). The results can be seen at http://dev.gentoo.org/~klondike/ebuildrank.txt and if there is interest I can try to automate the generation of the ranking daily (though I'd like infra's comments on that). We can put the script in qa-reports.g.o I think that kind of stats would be really easy to generate within the new p.g.o. Probably with a single database request. -- Corentin Chary http://xf.iksaif.net
Re: [gentoo-dev] [RFC] Initial python-r1.eclass distutils-r1.eclass
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 On 10/25/2012 10:13 PM, Michał Górny wrote: On Thu, 25 Oct 2012 20:55:37 +0200 hasufell hasuf...@gentoo.org wrote: Currently you seem to have focused more on distutils when writing python-r1 which makes this eclass a bit raw. Waiting for other developers to file feature requests instead of figuring out those yourself before they even consider porting their ebuild to your new eclasses seems like a questionable policy to me. They might not be too excited about it to start discussions and feature requests just to switch from an already working implementation. As you may have failed to notice, most of Python packages actually are using distutils. Thus, the core goal for distutils-r1/python-r1 was to correctly support those packages. Now that distutils is supported well, I can start thinking about supporting random hackish build systems. I will review redshift and give you my thoughts. Just note that your attitude is not motivating at all. I haven't killed any of your kitten or forced anyone to use python-r1. Most of you didn't even care to give a single word of feedback throughout the development process, so your position of 'this eclass doesn't give me shiny functions I want' is at least impolite. Sorry, I thought the main goal was to deprecate python.eclass at some (very distant) point. If that is true, then we need to support _all_ build systems using/related to python. I was just trying to say that you shouldn't wait for developers to point out all these cases. Best way would be to start converting exotic ebuilds and grep for packages that use python, but _not_ distutils. Otherwise I am missing the point why you created two eclasses instead of one (namely just distutils-r1). -BEGIN PGP SIGNATURE- Version: GnuPG v2.0.19 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/ iQEcBAEBAgAGBQJQitZQAAoJEFpvPKfnPDWzuOEIAJp9siMgIh4mdDP/kMNvCvpw jOJqML6ZMq9fEl5g8y7D46Vlw3cpQOUErh7fR7iNhN21EZsetNiAfi+s25+cnIWV XF/zrmdGxGJLqgJLwRI8sWwobCiQWpzQC+wJND6DDyCEk5NsJNMuCfFvgIO3l6YY Q6Amtn3QRwNGaZdCF6jWnSScqyJIK5x6ih6UVe99tgwPasNzDWxLesyr1LbkW4sB yohyQGN+JuSWlbOrM9BCs2M5VBFSMlnXTdwJqB4wxEY60FPsFQ33+5Mhx39Wvzd+ oDB+0NpqsVU6SZZ3K2hCr70T5M4j/CFGP1AaX5Z1nYoXxeJRkrat+95lcTCXeo0= =awW1 -END PGP SIGNATURE-
