Re: ad-interactive-form vs interactive-form

Re: ad-interactive-form vs interactive-form

[raman]

Hi Stefan,

This is a follow-up to the message I sent
 earlier this morning.

I spotted the cause of the problem  that I was hitting when using
interactive-form instead of ad-interactive-form:

interactive-form returns the autoload list for autoloaded functions,
ad-interactive-form used to return nil.

I therefore needed to add a check for (functionp (symbol-function sym))
before allowing Emacspeak to go off and auto-hack interactive functions.

Incidentaly, that corner of Emacspeak is a particularly hairy piece of
code that does:

0) Find functions that use an interactive spec for  declaring their
prompts.  1)    For those decls that dont use the minibuffer, perform
surgery to make those functions use the minibuffer.

The above was needed for a large number of functions until Emacs 21
--- since most  prompting resulting from interactive decls happened
directly through the C code in callint.c -- and consequently, Emacspeak
making functions like read-from-minibuffer or read-file-name was not
sufficient --- I needed to do the work  to perform run-time surgery to
functions that used interactive. As of emacs 21, I only need to do this
to commands that  use either 'c' or 'k' in their interactive spec --
would be nice to make those prompts go through the lisp layer as well --
in which case I could get rid of emacspeak-fix-interactive entirely.

Re: ad-interactive-form vs interactive-form

[Stefan Monnier]

> The above was needed for a large number of functions until Emacs 21
> --- since most  prompting resulting from interactive decls happened
> directly through the C code in callint.c -- and consequently, Emacspeak
> making functions like read-from-minibuffer or read-file-name was not
> sufficient --- I needed to do the work  to perform run-time surgery to
> functions that used interactive. As of emacs 21, I only need to do this
> to commands that  use either 'c' or 'k' in their interactive spec --
> would be nice to make those prompts go through the lisp layer as well --
> in which case I could get rid of emacspeak-fix-interactive entirely.

Yes I kind of saw that in the code.  I encourage you to post feature
requests via M-x report-emacs-bug.  I know there are often significant
resistance (including from me, of course), but even if the feature is
rejected, it's important for people to know that there's someone with this
need somewhere out there.

I for one would love to see a re-implementation of call-interactively all
done in Elisp, using an alist to map code letters to the corresponding code.
I have no time for that, sadly, and feel like it'd be a difficult sell on
top of it.


        Stefan

Re: ad-interactive-form vs interactive-form

[Richard Stallman]

    I for one would love to see a re-implementation of call-interactively all
    done in Elisp, using an alist to map code letters to the corresponding code.
    I have no time for that, sadly, and feel like it'd be a difficult sell on
    top of it.

If the result is clean, and doesn't cause a painful slowdown, I have
nothing against it.