This is the mail archive of the guile@sourceware.cygnus.com mailing list for the Guile project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

Not rushing the new module system


Normally in this project we reason that if there is an important need
and we have some implementation which fulfills it, we prefer adopting
the existing implementation before using resources to produce a
different one.

I probably don't need to say this, but I think the situation is
different concerning the module system.

We really need a strong and good module system to be a part of Guile's
infrastructure and it is important that this module system isn't
flawed like the previous two and that it is something which the
project can grow together with.

In my view, it will be very difficult to produce a good module system
which is at the same time backward compatible with the current one.
At any rate, switching to a new system will cause some amount of
trouble for Guile users.  Because of this, and because of how it looks
to people external to the project, this third system will be the final
one for quite some time to come.

Therefore, I think it's important that, even though it's urgent for
the project to get a good module system, we don't rush this
development too fast.  And I think it's extremely important that we
aren't going to use the (normally valid) argument that "the module
system I'm proposing exists while yours are still only talk".

Right now, we are only collecting ideas and references.  It might look
as if we are trying to design a system with too much features, but we
aren't designing yet, we're just collecting ideas which are good to
consider.  (Often the simplest and most elegant designs result from a
lot of consideration.)  After some time, we'll start a discussion
where we can talk about priorities of the various ideas we've
collected, and, finally, we'll make a new attempt at a design.

I think it is essential that we already before the discussion about
priorities have had a chance to study other implementations, and read
some papers, otherwise we'll just waste time.  Therefore, I suggest
that we wait perhaps a couple of weeks or two before even commencing
the real discussion about priorities.

/mdj

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]