[CM] Pointers into snd-help source?

[Michael Scholz]

Hi, Charles!

> (snd-help 'update-transform-graph)

It is corrected now in current fth-cvs.  It was a bug in
fth_documentation_ref|set.

You can use FTH's help word:

    help update-transform-graph

help is a parsing word so it's not necessary to produce an XT first.
snd-help is both a parsing word and a normal word:

    "update-transform-graph" #t snd-help
    snd-help update-transform-graph

> Also, from a quick look at the source, I can see why providing these for
> FTH (or Ruby) is non-trivial. Have there been any discussions/thoughts
> about how to approach providing the "H_" strings defined in the code now
> that Snd is multi-lingual?

All help strings are accessible from snd-forth and Bill has provided
some examples for all three languages!  If you get the latest fth-cvs
you will see them too.  I have replaced fth-1.0.8 on SF with corrected
tarball.  But these changes are not in it, sorry.

I use this channel for answering your private mail too because
optonline's sendmail says "User unknown" to your address.

> 1) I use Fink, and there are two place where ./configure says that "the
> preprocessor and compiler disagree about the validity of a header file."
> This whizzes by so quickly I don't have a clear memory of which files
> they are. I'll try to track this down for you and provide more info.

I have two or three candidates in /usr/include and I ignore these
warnings.  If you compile in an Emacs shell you can scroll to read the
messages.

> 2) The "make install" process quits because my computer doesn't like
> your "install info" command arguments. As a result, I think none of the
> .fs, .fr, etc files get installed, and I had to manually copy them to
> their directories. Also manually installed the FTH info pages.

doc/Makefile uses `install-info --info-dir=dir --info-file=file'

What option is wrong or what options knows your install-info?  In the
meantime, remove the last line at target install in doc/Makefile or
better in doc/Makefile.in before ./configure or ./config.status.

> 3) So I think the emacs sub-process for FTH doesn't work? Is it possible
> to get it to work from the code in Snd 8.9 release, or is there a lot of
> work to do? It's of course a really nice feature, but not essential.

This depends not on Snd.  I use this mode regularly and it works.  You
must set inf-snd-forth-program-name (note the forth in the name) to
your Snd program name.  Here is a part of my initialization in
.emacs.el:

(set-default 'auto-mode-alist
	     (append '(("\\.\\(fs\\|fth\\|fsm\\|fr\\)$" . snd-forth-mode)
		       ("\\.snd_prefs_forth$"           . snd-forth-mode))
		       auto-mode-alist))

(autoload 'run-snd-forth   "inf-snd" "Start inferior Snd-Forth process" t)
(autoload 'snd-forth-mode  "inf-snd" "Load snd-forth-mode, derived from `forth-mode'" t)

(setq inf-snd-forth-program-name "snd-forth-xm")

The rest is optional, I think.

(setq inf-snd-working-directory  "~/.snd.d/")
(setq inf-snd-index-path         "/usr/gnu/cvs/snd/")
(setq inf-snd-prompt             "snd> ")

(add-hook 'inf-snd-forth-mode-hook
	  (lambda ()
	    (setq inf-snd-comint-line-end "\n\n")))
(add-hook 'snd-forth-mode-hook
 	  (lambda ()
	    (define-key (current-local-map) "\C-co" 'snd-send-buffer)
	    (define-key (current-local-map) "\C-cr" 'snd-send-region)
	    (define-key (current-local-map) "\C-ce" 'snd-send-definition)))

> 4) Would there be a reason why compiling xg.c takes an order of
> magnitude longer with FTH than with Guile? This compile step has always
> been long, but for some reason with FTH, it's now a couple hours on my
> iBook (1ghz G4).

Hours?  Here it compiles not much longer than with Ruby or Guile.  I
confess I have no idea what's to do here.

All the best,

Mike