On Sat, Apr 09, 2016 at 12:56:49PM +0200, Pjotr Prins wrote: > On Tue, Apr 05, 2016 at 02:46:42PM -0400, Leo Famulari wrote: > > On Tue, Apr 05, 2016 at 12:02:28PM +0200, Pjotr Prins wrote: > > > I received a nice reply from Joe Armstrong. Basically he agrees with > > > the idea of deterministic builds and hashes for content of source > > > files > > > > > > http://joearms.github.io/2015/03/12/The_web_of_names.html > > > > > > and will support us if we raise our issue on the mailing list. Leo, do > > > you want to have a go - essentially sending the E-mail I wrote to Joe? > > > I can do it if you want me to. > > > > I'm more than happy to test things, but I think it's best if you send > > the message since you are actually using Erlang. But like I said before, > > if it's too much of a burden, I'll do it. Please let me know! > > A quick update: the Erlang authors are going to fix this issue as can > be followed in > > http://erlang.org/pipermail/erlang-questions/2016-April/088721.html
Good news. The changelog for http://erlang.org/download/OTP-19.0-rc1.README says that OTP-13504 Application(s): compiler The compiler will no longer put the compilation date and time into BEAM files. That means that two BEAM files compiled on the same computer from the same source code and compilation options will be identical. Note: If you want to find out whether a BEAM file on disk is different from the loaded code, compared the MD5 value obtained from Mod:module_info(md5) with the MD5 value obtained from beam_lib:md5(BeamFileForMod)` This means that Erlang and Elixir may go into GNU Guix :) Pj.
