> Author: eelco > Date: Tue Jul 13 22:35:50 2010 > New Revision: 22592 > URL: https://svn.nixos.org/websvn/nix/?rev=22592&sc=1 > > Log: > * PHP updated to 5.2.13. Removed the "if version == ..." stuff, it's > bad practice. > - # choose a different version this way: > - # php.merge { version = "5.3.2"; } > - version = "5.2.11"; > + version = "5.2.13"; [..] > src = args.fetchurl { > url = "http://nl.php.net/get/php-${version}.tar.bz2/from/this/mirror"; > - md5 = if version == "5.3.2" then "46f500816125202c48a458d0133254a4" > - else if version == "5.2.11" then "286bf34630f5643c25ebcedfec5e0a09" > - else throw "set md5 sum of php source file" ; > + sha256 = "1nzaqwsxah7hc59931dk81n2fxyqaaa3yqyisj6akf9vbr9s4l1b"; > name = "php-${version}.tar.bz2"; > }; >
Hi Eelco, I added the newer version for good reason. XMLRPC calls with SSL didn't work. Maybe they do now. I didn't check. Can you elaborate why it is bad style allowing the user choosing one out of multiple versions? Do you want each user to override the version in configuration.nix and maybe in ~/.nixpkgs/config.nix? I don't see it. That's why I'm asking. Also your description is not accurate enough: > * PHP updated to 5.2.13. Removed the "if version == ..." stuff, it's > bad practice. Would having a switch like statement be less bad style? In other words: This is the use case why I sent the Nix patch about allowing arbitrary strings as attr names so that the "version ==" doesn't have to be repeated. So if the choice which was implemented by "if" was represented by kind of case would it be acceptable by you then? The example was given here: http://thread.gmane.org/gmane.linux.distributions.nixos/4526 Macr Weber _______________________________________________ nix-dev mailing list [email protected] https://mail.cs.uu.nl/mailman/listinfo/nix-dev
