Marc Weber wrote: > Excerpts from Yury G. Kudryashov's message of Thu Dec 23 09:07:36 +0100 > 2010: >> > If I compare this to: >> > - introduce NIX_SCONS_PATH (- because complexity increases much more) >> > - users have to look it up >> Buildenv can set NIX_SCONS_PATH to NIX_PROFILES/bin:NIX_PROFILES/sbin. > About derivations you're right. You also have to think about installing > scons into your user profile. > > Then you have to look it up. Then you try to write your first scons > build script and the result will be "gcc not found". No since NIX_SCONS_PATH is set to NIX_PROFILES/bin:NIX_PROFILES/sbin. > > You should try to find a use case where my patch does any real harm to > anybody. The reason why it was written is clear. I have nothing against your patch (btw, I don't use scons). Just some (not all!) of your arguments against NIX_SCONS_PATH approach are wrong. > > If I read the thread only "a compromise" such as the new env var always > looks fine. But its not the most simple solution. Thus its bad. (IMHO). > > If we started agreeing on "compromises" for wrong reasons it would be > bad. Since I don't use scons, I don't want to make a decision here.
_______________________________________________ nix-dev mailing list [email protected] https://mail.cs.uu.nl/mailman/listinfo/nix-dev
