[Nix-dev] Re: [Nix-commits] SVN commit: nix - 16100 - MarcWeber - in nixpkgs/trunk/pkgs: applications/editors/eclipse applications/graphics/cinepaint applications/graphics/gimp/plugins build-support/buildenv development/interpreters/ruby top-level

Marc Weber marco-oweber at gmx.de
Tue Jun 30 18:33:12 CEST 2009


> Was it your intent to commit all these debug statements in buildenv/builder.pl?
reverted. of course not. sorry. That was an accident.

> > +  # TODO don't resolve 302 redirects but make nix resolve in fetchurl and
> > +  # nix-prefetch-url. This should be done on stdenv-updates.
> 
> Changes to fetchurl are harmless so they can also be done on the trunk.
See my other post (goto "Eg put"):
http://article.gmane.org/gmane.linux.distributions.nixos/2228

Changing an url will cause another store path. I think the same applies
for the builder script then?
 
> > +  # ================ START automatically generated code ================
> 
> Isn't it better to have the generated code in a separate file?
Maybe. Don't think it matters enough to fix it. I don't mind you fixing
it if matters to you.

> 
> > +
> > +           # WARNING: automatically generated CODE
> > +           # This section has been generated by gem nix sup rubygems-update
> > +           # the gem nix command has been added by a nix patch to ruby gems
> > +        
> > +    rubygems_update_1_3_4 = rubyDerivation {
> > +       name = "rubygems-update-1.3.4"; # full_name
> > +       nameNoVersion = "rubygems-update";
> 
> Why not have a "name" attribute (without the version) and a "version" attribute?
Laziness. I needed that value so I added it. Using both would have
required me fiddling with substr functions ?
This ruby gems stuff is still a proof of concept. Am glad that it alrady
works very well.
 
> > +         description = "RubyGems is a package management framework for Ruby.
> > +
> > +  This gem is an update for the RubyGems software.  You must have an
> > +  "; # cut to 120 chars
> 
> meta.description should not contain newlines since it screws up nix-env -q output.

Which is the best way to fix this?
making nix-env -q ignoring newlines (or replacing them by spaces)
or fix nix derivations?

Marc Weber



More information about the nix-dev mailing list