On Tue, Feb 23, 2016 at 12:33 PM, Anil Madhavapeddy <a...@recoil.org> wrote: > On 23 Feb 2016, at 12:28, David Sheets <she...@alum.mit.edu> wrote: >> >> The Docker containers described by >> https://github.com/ocaml/opam-dockerfiles somehow set the OPAMSWITCH >> environment variable. I haven't been able to find where or why this >> variable has been set and it causes many opam commands to be more >> heavyweight than necessary which makes interactive use annoying. For >> instance, after switching compilers, you must run something like >> >> ``` >> eval `opam config env --switch=system` >> ``` >> >> if OPAMSWITCH is set. When is OPAMSWITCH necessary? Why? What sets it? > > One possibility is that the entrypoint to the container passes the > command through "opam config exec --", which ensures that the local > OPAM context is available when you run the command.
Ah, yes, that does look like the culprit. I've submitted <https://github.com/ocaml/opam/pull/2444> to change this behavior. > This isn't really designed to support compiler switching, now that > I think about it :-) Perhaps we should rejig the container layout > to have precisely one switch per container? Do you just want to use > the system switch in your usecase? No, I would really like to be able to switch to, e.g., 4.02.3+bytecode-only which Jeremy added recently at Daniel's request. I don't think it makes sense to have tags for all compiler switches in the universe but I agree that in automation use cases there should generally be a single switch in each container. David _______________________________________________ opam-devel mailing list opam-devel@lists.ocaml.org http://lists.ocaml.org/listinfo/opam-devel
