[Nix-dev] Announcing nix-buffer, nix-shell for emacs

[zimbatm]

Interesting. Would it also work if you set "coq-prog-name" to "coqtop" and
let emacs resolve the program name from the PATH?

On Tue, 6 Sep 2016 at 10:39 Shea Levy <shea at shealevy.com> wrote:

> Hmm, I think we need more than just environment variables. See for
> example my recent commit[1] to nixpkgs, where in addition to setting up
> process-environment and exec-path I also set, when appropriate, several
> proofgeneral variables like coq-prog-name and even call some
> proofgeneral functions.
>
> ~Shea
>
> [1]:
> https://github.com/NixOS/nixpkgs/commit/b04cdae902398aa76fb81a0470e18fd527c6ba3d
> zimbatm <zimbatm at zimbatm.com> writes:
>
> > Hi Shea,
> >
> > did you look at direnv[1] yet? I have a feeling that it would mix with
> > emacs pretty well and implement all that you need in a more generic way
> > (which is a pro and con).
> >
> > Essentially what I do in my projects is use the following .envrc:
> > ```bash
> > use nix
> > ```
> > The `use nix` directive calls nix-shell under the hood so then the usual
> > default.nix and shell.nix are loaded.
> >
> > In emacs the plugin would need to call `direnv export json` on
> enter/leave
> > of a context. This gives you a diff of environment variables to apply.
> > There is a bit more complexity when implementing the security framework
> but
> > that's about it.
> >
> > I'm just curious if you thought of that approach, keep up the good work!
> >
> > Cheers,
> > z
> >
> > [1]: http://direnv.net/
> >
> > On Mon, 5 Sep 2016 at 17:11 Shea Levy <shea at shealevy.com> wrote:
> >
> >> Sure! In my .emacs, I have:
> >>
> >> > (defun nix-buffer-find-file-hook ()
> >> >   (when (not (file-remote-p (buffer-file-name)))
> >> >     (nix-buffer)))
> >> > (add-hook 'find-file-hook 'nix-buffer-find-file-hook)
> >>
> >> I have coq 8.5pl1 installed in my configuration.nix, but I have a coq
> >> project that requires 8.4pl6, so I have in that project's root directory
> >> dir-locals.nix:
> >>
> >> > let pkgs = import <nixpkgs> {}; in
> >> >   pkgs.nixBufferBuilders.withPackages [ pkgs.coq ]
> >>
> >> (note that this relies on a recent nixpkgs commit[1]). Then whenever I
> >> open a .v file in the project, ProofGeneral uses coq 8.4 rather than
> >> 8.5.
> >>
> >> If I wanted, my dir-locals.nix could take a 'root' argument, which would
> >> point to the file I opened.
> >>
> >> ~Shea
> >>
> >> [1]:
> >>
> https://github.com/NixOS/nixpkgs/commit/05c132486d8cfae600bbfe8c9ac5d799b298afed
> >>
> >> Bas van Dijk <v.dijk.bas at gmail.com> writes:
> >>
> >> > Hi Shea,
> >> >
> >> > As a Nix and Emacs user this looks very interesting. I didn't know
> about
> >> > Buffer-Local Variables before.
> >> >
> >> > Can you share some applications of nix-buffer / how you are using it?
> >> >
> >> > Thanks!
> >> >
> >> > Bas
> >> >
> >> > On 5 September 2016 at 08:06, Shea Levy <shea at shealevy.com> wrote:
> >> >
> >> >> Hi all,
> >> >>
> >> >> I've just pushed the initial version of nix-buffer[1], and opened a
> PR
> >> >> for adding it to melpa[2]. The description of the 'nix-buffer-enter'
> >> >> command this library defines:
> >> >>
> >> >> > Set up the buffer according to the directory-local nix expression.
> >> >> >
> >> >> > Looks for dir-locals.nix upward from the current directory.  If
> found,
> >> >> > builds the derivation defined there with the 'root' arg set to the
> >> >> > current buffer file name or directory and evaluates the resulting
> >> >> > elisp if safe to do so.
> >> >> >
> >> >> > Because in practice dir-locals.nix will always want to do things
> that
> >> >> > are unsafe in dir-locals.el (e.g. append to 'exec-path'), we don't
> >> >> > reuse that mechanism and instead just load the file as elisp.
> Because
> >> >> > this allows arbitrary code execution, the first time we're asked to
> >> >> > load a particular store path we query the user to verify if it's
> safe
> >> >> > to load beforehand.
> >> >> >
> >> >> > The Lisp code generated by dir-locals.nix should limit itself to
> >> >> > modifying buffer-local variables, but there is no actual
> enforcement
> >> >> > of this.  'setq-local' is your friend.
> >> >>
> >> >> It may be useful to advise the 'normal-mode' function to call
> >> >> 'nix-buffer-enter' beforehand for local files, so that all visited
> files
> >> >> have their environments set up before modes are loaded.
> >> >>
> >> >> Note that this is my first elisp package, so feedback definitely
> >> >> welcome!
> >> >>
> >> >> In addition to the elisp work mentioned in the TODO[3], it would be
> >> >> useful to add some functions to nixpkgs to build elisp files,
> >> >> especially:
> >> >>
> >> >> * A function taking a derivation and building an elisp file that
> sets up
> >> >>   the buffer environment analogous to the build environment
> specified by
> >> >>   that derivation
> >> >> * A function taking a list of packages and building an elisp file
> that
> >> >>   sets up the buffer environment with all of those packages available
> >> >>   (in PATH, in the emacs load path, etc.).
> >> >>
> >> >> Happy hacking!
> >> >>
> >> >> ~Shea
> >> >>
> >> >> [1]: https://github.com/shlevy/nix-buffer/tree/v1.0
> >> >> [2]: https://github.com/melpa/melpa/pull/4204
> >> >> [3]: https://github.com/shlevy/nix-buffer/blob/v1.0/TODO.md
> >> >>
> >> >> _______________________________________________
> >> >> nix-dev mailing list
> >> >> nix-dev at lists.science.uu.nl
> >> >> http://lists.science.uu.nl/mailman/listinfo/nix-dev
> >> >>
> >> >>
> >> _______________________________________________
> >> nix-dev mailing list
> >> nix-dev at lists.science.uu.nl
> >> http://lists.science.uu.nl/mailman/listinfo/nix-dev
> >>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.science.uu.nl/pipermail/nix-dev/attachments/20160907/991049f1/attachment.html>