Re: Intermediate review for SF #1708292

[Jeffrey H. <jc...@ho...>]

Bill Wohler <wo...@ne...> wrote:

> Jeffrey Honig <jc...@ho...> wrote:
> 
> > So, what I came up with is the requirement to pass a syntax table to
> > mh-regexp-in-field-p and mh-modify-header-field.  Specifying nil will
> > cause mh-modify-header-field to make an assumption based on the field
> > name.  Does this sound like overkill?  Maybe all we need is
> > mh-modify-header-field to make a best guess?
> 
> Remind me how syntax tables are used programmatically. What call
> specifically is the one that needs it? Is it the re-search-forward call
> in mh-regexp-in-field-p? Passing a syntax-table around as an argument
> has a bad smell to me also. Can you just override a local variable with
> let just before it is needed?

Syntax tables are used by regexp call.  So yes, we would need to wrap
the re-search-forward call (and only that call).  It is necessary to
use (set-syntax-table) to change the table and (syntax-table) to get the
current one.  So the re-search-forward would need an (unwind-protect
(save-excursion)) wrapper.

> If the syntax table is only really needed in mh-regexp-in-field-p, the
> lowest-level function, then perhaps we can put the syntax table smarts
> in there. That way, we don't have to pass around a syntax table, nor do
> we have to figure out what syntax table we need everywhere we call
> mh-modify-header-field or mh-regexp-in-field-p.

After sleeping on it, I think that is best.  We can also provide a
global variable that can be set with a let as a hint.

Thanks!

Jeff

-- 
Jeffrey C. Honig <jc...@ho...>
http://www.honig.net/jch
GnuPG ID:14E29E13 <http://www.honig.net/jch/key.shtml>