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

[Marius Vollmer]

Rob Browning <address@hidden> writes:

> But what is it that you want out of top-repl?

I don't know.  <shrug>

>   (begin
>     (top-repl-init)    ;; steps 1-2
>     (load-user-init)   ;; code generated from args
>     ...                ;; more code generated from args
>     (top-repl-execute) ;; steps 3-5
> 
> This split is necessary because the code generated by the args might
> depend on the envt being set up properly by steps 1-2.

Do we want to support his?

> Actually, it's a little more complicated than that since (for
> example) after we finish parsing the args, we have to look and see
> if we're going to end up in interactive mode (i.e. was -c or -s
> specified), and if not then step 1 is mostly skipped.

That's what I don't like about it, the conditionality.  The main
reason why the additional modules are not loaded unconditionally now
is that they slow down startup and then might not be needed.  For the
repl, they are loaded since they provide convenient bindings for
interactive programming, like `help' and `apropos'.

> In the end perhaps we should have two public repl functions, one which
> just publishes the "normal repl".  If you call this function, you get
> the same repl that you'd get when you run guile interactively.  The
> other one would be a lower level repl function that mostly just did
> step 5 for those people who want/need to set things up in a more
> customized way.  Step 5 is really where the trickier bits are.

Yes.