On 13 April 2011 11:36, Marijn <hkBst@g.o> wrote:
> On 04/13/11 03:45, Imran Rafique wrote:
>> Will do. But there's one issue. I needed to replace /usr/bin/mzscheme
>> with a symlink to /usr/bin/racket.
>>
>> Racket-5.1 (from this overlay) installs a /usr/bin/mzscheme binary in
>> addition to /usr/bin/racket. Seems like some useless leftover artifact
>> ('racket' provides the repl, 'mzscheme' segfaults)
>
> I wasn't aware, but I just checked and with the live ebuild that I have
> locally I indeed still get that file too and it does:
>
> $ mzscheme
> read failed to read all 31 bytes from file /usr/bin/mzscheme
> Aborted

Yup, same here.

>> Vim-7.3 was released when racket was still plt-scheme, and it
>> specifically looked for the mzscheme binary. vim-7.3.049 updated the
>> scheme support to work with racket, but vim still looks for
>> '/usr/bin/mzscheme' and uses it (sure, thats an upstream issue, but
>> its what we have right now).
>>
>> Replacing /usr/bin/mzscheme with a symlink to /usr/bin/racket sorted
>> that issue out, but this really needs to be something which the racket
>> ebuild itself resolves (whats that mzscheme binary for, anyway?)
>
> I'll talk to upstream about the mzscheme executable.

I just saw your mail on racket-dev. I was too lazy to do that - thanks :)

> How hard would it be to fix vim to look for racket instead of mzscheme?

Hmm, not sure. I had a quick look at patch 049 (which updated from plt-scheme
to racket) but didn't see anything obvious. There are other issues with how
racket is embedded. For example, the scheme snippet in configure, which detects
what racket's install prefix is, doesn't work. Its missing the '#lang scheme'
or a (module ...) form (I'll mention that to upstream).

Thats easily handled by my vim.eclass by bypassing this detection of racket's
prefix, by passing "--with-plthome=/usr/".

One quick way of fixing the mzscheme problem from our end (for now), would be
for the racket ebuild to just replace the mzscheme binary with a symlink to
racket, and install that. In keeping with the gentoo spirit of quick local
fixes while waiting for upstream to fix the problem :)

--
Regards,
       Imran Rafique