On Sun, 12 Feb 2012 14:09:57 +0200, Jani Nikula <jani@nikula.org> wrote:
> On Fri, 10 Feb 2012 10:05:46 +0000, David Edmondson <dme@dme.org> wrote:
> > On Thu,  9 Feb 2012 14:46:02 +0000, Jani Nikula <jani@nikula.org> wrote:

[...]

> > > +	      (if (or (equal notmuch-show-all-multipart/alternative-parts t)
> > > +		      (string= chosen-type inner-type)
> > > +		      (and
> > > +		       notmuch-show-all-multipart/alternative-parts
> > > +		       (equal (string-match-p (mapconcat (lambda (s) (format "^%s$" s)) notmuch-show-all-multipart/alternative-parts "\\|") inner-type) 0)))
> > 
> > This is quite messy. How about we add a general helper to "notmuch-lib.el":
> > 
> > (defun notmuch-string-match-list-p (regexps string)
> >   (loop for regexp in regexps
> > 	if (string-match-p regexp string)
> > 	return t))
> > 
> > and then write:
> > 
> >     (if (or (string= chosen-type inner-type)
> >             (equal notmuch-show-all-multipart/alternative-parts t)
> >             (notmuch-string-match-list-p notmuch-show-all-multipart/alternative-parts inner-type))
> >        ...
> 
> Hmm. I wrapped the regexps between ^ and $ to not match substrings. I
> think that's less surprising and better than having the user add them,
> and the user can still use e.g. ".*foo.*" to explicitly match a
> substring.
> 
> If this is fixed in notmuch-string-match-list-p then it's not general
> anymore. But does adding a mapcar to wrap the regexps here make this
> messy again...? What if notmuch-string-match-list-p were just a local
> specific helper for clarity?

Something like this could work...

 (defun notmuch-string-match-list-p (regexps string)
   (loop for regexp in regexps
 	if (string-match-p (concat "^" regexp "$") string)
 	return t))

But if you-re going to pre-anchor these regexps then it is better to be
documented clearly as the final 'regexp' works in non-standard way.

> > 
> > ?

Tomi