User-Agent: Mozilla/5.0 (Windows NT 6.1; WOW64; rv:24.0) Gecko/20100101
            Thunderbird/24.5.0
MIME-Version: 1.0
References: <538C959C.3060904@morningstar2.co.uk>
            <F7E66580-32F8-4385-A765-942E4303B7F6@gmail.com>
Content-Type: text/plain; charset=windows-1252
Message-ID:  <538D9231.1030807@morningstar2.co.uk>
Date:         Tue, 3 Jun 2014 10:15:29 +0100
Reply-To: Mailing list for the LaTeX3 project
              <LATEX-L@LISTSERV.UNI-HEIDELBERG.DE>
Sender: Mailing list for the LaTeX3 project <LATEX-L@LISTSERV.UNI-HEIDELBERG.DE>
From: Joseph Wright <joseph.wright@MORNINGSTAR2.CO.UK>
Subject: Re: 'Preseting' keys
To: LATEX-L@LISTSERV.UNI-HEIDELBERG.DE
In-Reply-To:  <F7E66580-32F8-4385-A765-942E4303B7F6@gmail.com>
Precedence: list
Content-Transfer-Encoding: quoted-printable
Envelope-To: <rainer.schoepf@GMX.NET>
Status: R

On 03/06/2014 00:17, Will Robertson wrote:
> Hi Joseph,
>=20
> I think the name and the functionality are sensible.

Cool :-)

> The first case is what I do in fontspec (but there=92s lots I do in fon=
tspec that I probably shouldn't).
> I suspect that I=92d need to play around with some interfaces to figure=
 out how this would change things.
> Thinking along these lines:
>=20
>=20
>>    \cs_new_protected:Npn \foo #1#2 % #1 =3D keys
>>      {
>>        \keys_define:nn { mymod }
>>          {
>>            key-a .preset:n =3D new-value ,
>>          }
>>       \keys_set:nn { mymod }
>>          {
>>            #1
>>          }
>=20
>=20
> One thing that becomes a bit harder is unsetting the preset; if I then =
need to run \keys_set:nn again the preset would either need to be grouped=
 or explicitly disabled.
>=20
> (E.g., when setting Letters=3DSmallCaps for the small caps font before =
moving on to another font shape.)
>=20
> So at least in my use case, and only thinking very briefly about it, I=92=
m not sure if this interface would be a advantage for fontspec =97 but th=
at=92s not to say that it=92s not a bad idea, nor either that fontspec co=
uldn't be re-jigged to be a bit more sensible in such matters.
>=20
> Cheers,
> Will

I'm seeing offering a 'preset' concept as a convenience/performance
boost, but not necessarily trying to cover everything. The scenario you
outline for fontspec is probably tricky to cover without some dedicated
code, although I guess one might have

  key-a .unset:

or

  key-a .skip_preset:

I think a more common issue is how presetting interacts with filtering
or grouping. Not sure at present, but I suspect the most sensible
outcome is that filters/groups apply to preset keys such that only
'active' keys (that potentially could be set) are preset.
--=20
Joseph Wright