Hi, yes, this does make sense, and many thanks for writing this down in detail. - I will follow your steps (and develop my own routine).
And computers are getting faster after all (to some extent still): My 53sec for pulling were on an 4 core i5 (with 16G RAM) - this e-mail I am writing on a 8 core i7 w/ 16G RAM as well - and I might want use it for guix at some point (and that should accelerate things then). All the best, -Andreas zimoun <zimon.touto...@gmail.com> writes: > Hi, > > For developing, my workflow looks like: > > 0 git clone http://git.a-rx.info/channel/ /tmp/channel > 1 edit /tmp/channel/some-file.scm > 2 guix build -L /tmp/channel ghc-filepatter > 3 loop to 1 until I am satisfied > 4 git -C /tmp/channel commit ... > > then it depends on my mood to have http://git.a-rx.info/channel/ > listed in my default channels file ~/.config/guix/channels.scm. Well, > when I am tired to type '-L path/to/channel' then I add it and I run: > > guix pull > > which allows then to simply do > > guix package -i ghc-filepattern -p path/to/profile > > Does it make sense? > > About "guix pull", yes it should appear slow. But you cannot do more > than wait... although the situation is mitigated by [1]. (Optimizing > it is not as straightforward as it looks like at first because the > bottleneck are at various stages. :-) For instance, > <http://issues.guix.gnu.org/39258#124>.) > > Have you seen 'channel-with-substitutes-available' [1]? > > 1: > <http://guix.gnu.org/manual/devel/en/html_node/Channels-with-Substitutes.html#Channels-with-Substitutes> > > All the best, > simon