Re: Discrepancy in definition/use of match-data?

[David Kastrup]

"Stephen J. Turnbull" <address@hidden> writes:

> >>>>> "rms" == Richard Stallman <address@hidden> writes:
> 
>     rms> sjt writes:
> 
>     On the other hand, pretty much any time any Lisp code intervenes
>     between the return of the matching function and entry to the match
>     data access there is an opportunity for a hook or handler to be
>     called.
> 
>     rms> That is rather an exaggeration.
> 
> Exaggeration, sure.  But it contains a kernel of truth.

Those things that run in hooks/handlers/whatever need to get wrapped
in save-match-data, anyway.

>     rms> Most of the Lisp functions described in the manual cannot
>     rms> call any hook.
> 
> True, but not relevant unless you know which ones they are.

> Offhand, I don't, and the docstrings/source comments are less than
> 100% reliable.

Where the docstrings are not reliable, they need to get amended,
obviously.

> Anyway, I've implemented a last-match-succeeded flag (for debug
> usage only) and check it in match_limits and Fmatch_data (when
> XEmacs is configured for error-checking).  If we catch anything that
> looks relevant to GNU Emacs I'll report it here.

Anyway, the args-out-of-range error that now gets thrown gives a
basically completely irrelevant second value (with regard to
identifying the error.  Actually, the first value is also irrelevant
unless it is negative).  Could one replace that second value with the
function causing the error instead, or would that be inconsistent with
any known error handlers?

Or should one throw a different error altogether in the case where
match-data is called without a valid match ever having been done?
Which one?  As I said already, I have no clue about the error handling
conventions.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum