[racket-dev] Installing subsets of Racket

Hi Laurent,

I think that the solution to this are "binary" builds.... versions of
a package that only have the bytecode and documentation.

We're a bit behind on binary builds, because when they were discussed
for the main repository [1] they were rejected. I hope to be able to
still provide them for ring-0 packages through the results of DrDr
running tests (and thus compiling) on those packages, but it's in the
future.

The result would be that when you installed a package in "binary"
form, you would only get the "deps" and not the "build-deps". (And
you'd probably get those in binary form too.)

Jay

1. http://www.mail-archive.com/dev@racket-lang.org/msg08879.html

On Mon, Sep 16, 2013 at 2:32 AM, Laurent <laurent.orseau at gmail.com> wrote:
> Hi,
>
> (this is not a complain, just an inquiry)
>
> While installing Racket on a small server, I wanted to avoid installing gui
> and doc related libraries.
> The minimal install was great!
>
> Then I wanted to install a package of my own (the aptly named "bazaar"),
> which requires "images" and other gui libs (which I actually would not use
> on the server), among other things, but no doc
>
> But the "images" package draws racket-doc and gui-doc dependencies, which in
> turn draws practically all of Racket. And it then takes a much longer time
> for `raco setup` to do its job that I had hoped for.
>
> Certainly, this can be resolved by splitting "images" and "bazaar" into lib,
> gui and docs packages, but I foresee another problem:
> It's difficult to enforce such a split for third-party libraries, as it puts
> the burden on the user.
> And the first package like that to be installed will again draw all of
> Racket dependencies.
>
> This is probably not a trivial matter, but what can be done about this?
>
> My dream would be that gui and doc dependencies are never triggered, without
> preventing the packages I actually use to be downloaded, but I don't know
> how this could actually be ensured without a good amount of magic.
>
> Merely preventing downloads does not sound like a good option though.
>
> I bet you've already discussed this far and wide, so are there any plans?
>
> Laurent
>
> _________________________
>   Racket Developers list:
>   http://lists.racket-lang.org/dev
>



-- 
Jay McCarthy <jay at cs.byu.edu>
Assistant Professor / Brigham Young University
http://faculty.cs.byu.edu/~jay

"The glory of God is Intelligence" - D&C 93