Re: Recent guile-user related reversion in boot-9.scm

[Rob Browning]

Marius Vollmer <address@hidden> writes:

> Phfrr, don't know.  I'm using it in my simulator, which has its own
> way to set up the environment for the user, passing the buck to
> top-repl in the end.  So I would say, yes, there should be a
> advertised way to start the standard repl, but maybe it should not
> be top-repl but rather something more consciously designed.  For
> example (related to something you say in another post), there should
> be a way to not switch to (guile-user) for the repl, etc.

OK.  Well, I'm working on that code right now, and it's just been
redesigned anyhow, so it's a good time for changes if we can figure
out what we want.

Maybe something as simple as (top-repl . module) where module, if
present indicates a module other than (guile-user).

> I'd say, nothing should be in the script environment except (guile),
> which is hard to avoid anyway.  In the interactive encironment, I
> don't know... probably what is currently in there: regex, threads,
> session, debug, autoloaded debugger.

OK.  I'll restructure things along those lines, and I'll continue to
provide top-repl publically.

> However, I'm also growing sympathetic towards meta commands in the
> repl.  I don't know exactly what I was opposed to in the past, but now
> I can see the value of such commands, like being able to type

I don't have any strong feelings about the meta commands either way,
as long as there's still a "normal" way to do the same things, and as
long as it doesn't mess up any other important semantics, or make any
guarantees that a compiler (for example) would have a hard time with.
I presume that's not relevant since it sounds like you mean that
meta-commands should only work from an interactive prompt.

-- 
Rob Browning <address@hidden> PGP=E80E0D04F521A094 532B97F5D64E3930