[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: proposal: new ocaml dirs schema



Sven Luther <luther@dpt-info.u-strasbg.fr> writes:

> Ok, ...
>
> But how do we move the symlinks around ? Do we use a plain
> alternative/diversion thingy (need to recheck on those) or do we do
> something more fancy ? How do we force the symlinks to be ever on the
> last installed package ? I don't believe we need them really, since i
> don't think we want a trully // install, but something more asymetric,
> where the last released version of ocaml (3.06 today) is the one used,
> the other being more an additional service for peoples who really need
> it, and are ready to do a bit more work to make it work.

No alternative, no diversion.

A dummy package must provide symlinks to the correct binaries.
This package has the version of the standard Python.

>
> I can even provide a Makefile.3.04 or something such file, which would
> define the OCAMLC, OCAMLOP, etc to point to the ocaml 3.04 binaries,
> without real need for a symlink thingy.
>
>> > But then, like i said, we have time to speak about it, and convince us
>> > all that we did choose the right solution.
>> 
>>   I'd like to see something implemented before telling anything ;-)
>>   But again, I'm against any package duplication. 
>
> Ok, the next version of ocaml i will provide, (well apart from the
> bugfixe ones maybe), will move the ocamllib dir, and no more. Mmm, maybe
> i should not really wait, and upload it. This would need a rebuild of
> all the ocaml libraries though.

Err, this is an experiment, so it must be uploaded either to experimental
or on a website.

>
> And then, as an example, i will package ocaml-cvs, and you can look and
> see how it works. There will, at first, not be any kind of libraries for
> this one, so it will be easier. Then, if we are satisfied with it, i can
> rebuild ocaml 3.04, and you can play some with the libraries to see if
> you like it.

IMO, there is no need to have ocaml-cvs and ocaml installed at the same
time.

-- 
Jérôme Marant

http://marant.org



Reply to: