Re: any news on 1.8 and slib?

[Rob Browning]

Neil Jerram <address@hidden> writes:

>> With respect to SLIB, as mentioned, I was talking to Aubrey Jaffer,
>> and at the time it sounded like we might want to follow a different
>> approach in 1.8, one where each SLIB feature would have its own Guile
>> module with its own exports.  I didn't get far enough to verify, but
>> it sounded like SLIB might support enough introspection to allow us to
>> automate the modularization.
>
> Presumably it would still work for a program to use SLIB's require
> mechanism instead?

I think so.  For example, (require 'format) might just expand into
something like (use-modules (slib format)).  I assume that people
shouldn't call use-modules directly (aside from the initial
(use-modules (ice-9 slib))).

>> So at this point, if someone else wants to investigate, that would be
>> great.  Also, there's no reason that we have to pursue the more
>> ambitious changes.  For now, if no one has time to investigate the
>> modular approach, we could attempt to do something more like what we
>> did in 1.6.
>
> Can you remind us what the issue is?  (A pointer to an existing email
> would do.)

The quickest way to pursue an approach like the one in 1.6 would be to
grab the latest SLIB and then just change the slib.scm in 1.8 to be

  (define-module (ice-9 slib))
  (load-from-path "slib/guile.init")

Then look through guile.init, test SLIB, see what breaks, and propose
any necessary changes to SLIB's guile.init upstream.

When I started doing something similar a while back, I noticed a
number of things in guile.init that I wasn't sure about, so I asked
Aubrey Jaffer about them, and that's what led to the discussion about
modularization.

If anyone wants to attempt the modularization, I can look back and try
to summarize my conclusions there as well.

-- 
Rob Browning
rlb @defaultvalue.org and @debian.org; previously @cs.utexas.edu
GPG starting 2002-11-03 = 14DD 432F AE39 534D B592  F9A0 25C8 D377 8C7E 73A4