Re: [Chicken-hackers] new CHICKEN 5 egg installation

[felix . winkelmann]

>  Exciting news! What would it take to get deployment working? In the
> past, I found it immensely useful to be able to distribute binary,
> particularly with Windows and Mac OS X users who are not likely to
> have a compiler installed, or a readily installable Chicken binary
> distribution.

Deployment is indeed useful, but the complexity of supporting all
different OSes should perhaps be extracted into a separate tool
that can be maintained at a quicker pace than the core system.

I am repeatedly frustrated by overengineered tools (go, the Haskell
Stack, OPAM, etc.) that try to handle all sorts of things and in the
end break horribly because that options are so many. "csc" can
still support the lowlevel things (like "-deployed"), but wrapping
up, copying libs and so on might be better handled by another egg
or by detailed documentation in the CHCIKEN wiki.

> Related to that, would just fetching the sources for an
> egg and all its dependencies be possible? It was a lot easier to
> create binary deployments that way.

Yes, that is already possible ("-retrieve -recursive"), I think.

>
>   As for getting rid of install prefixes, I also find it useful to be
> able to install eggs in a user's home directory, as opposed to in /usr
> or /usr/local. But if you find it so annoying, should we try to
> brainstorm of a less confusing scheme for installing in non-standard
> locations?

Yes, we need to think more about this.

> I think it is sensible to abort the operation if a test in a dependency fails.

Ok, noted.

>
> Is there a replacement of matchable in Chicken 5, if syntax-only
> extensions are not permitted?

Matchable is still supported, what is not permitted is the syntax-only
indicator. Currently this means that a small piece of useless runtime code
(matchable.so) will be loaded, if though it is not needed.

Thanks for the suggestions, Ivan!


felix