[tex-live] [lltx] Updating TeXLive 2010 using tlgmr
wspr81 at gmail.com
Wed Sep 21 11:36:12 CEST 2011
On 16/09/2011, at 7:21 AM, Reinhard Kotucha wrote:
> On 2011-09-16 at 03:50:47 +0930, Will Robertson wrote:
>> On 16/09/2011, at 12:00 AM, Khaled Hosny <khaledhosny at eglug.org> wrote:
>>> I'm not sure what is the issue here either, but I think it would
>>> be good to not try to load fonts that is never used in the
>>> document, e.g. if I use fontspec and set a roman font and don't
>>> have any text using sans or typewriter fonts then why I need LM
>>> or any other font that I'm not using at all?
>> Because users first try to load the package without selecting a
>> font and wonder why their accents don't work.
> don't be worried too much about the missing accents. This is a
> special case.
I seem to remember this came from actual user bug reports, not a hypothetical situation.
So not going to happen. People *often* write minimal examples that look like
% something that may or may not involve accents breaking
>> I might change fontspec to only load LM if no other font has been
>> selected in the preamble.
> Yes, that would be great.
Looking quickly into this I've decided *not* to do it at this stage. The problem is that I need to change encodings and I can't really see the point to complicate the code: I think it'd have to be something like
* load unicode font encoding (EU1 or EU2 at present)
* select OT1 again
* after the default fonts are set, select EU1/EU2
* at begin document check if the above has happened
- if not, work out which of rm/sf/tt have been set and those that haven't been use LM
- then select EU1/EU2
Versus, at present:
* select LM
* load EU1/EU2
If someone can make a good argument that loading LM by default is a bad idea for some reason, I'm happy to reassess this. But making the code uglier to make the internal process cleaner isn't a great idea in my eyes. Let me know if I'm wrong :)
More information about the tex-live