Re: [NTG-context] Latin Modern bug in lm-texnansi-os.enc

2007-05-25 Thread bop
Hi,

 On the ConTeXT mailing list, Idris posted some messages because he was
 trying to use oldstyle numerals in the latest Latin Modern release,
 and was experiencing two pretty big problems:
 
* really bad kerning around the oldstyle numerals

We would be grateful if we could get more details (which font, which tfm,
which
encoding, pdftex or tex/dvips, on the screen or on the paper, etc.) My first
guess is that proportional figures were used with tabular metrics...

* pdftex generated errors for the typewriter fonts, complaining
  glyph `one.oldstyle' undefined etc.

Yeah... This is the result of our decision to comply with the currently
obeying
naming conventions (better to say: fashion). The oldstyle figures that appear
in
ts1 (text companion) encoding are actually *.taboldstyle; the names *.oldstyle
are reserved for proportional glyphs. We included also proportional oldstyle
figures in LMs (and in the TeX Gyre fonts).

 The implementation of 'automatic' oldstyle numbers in ConTeXt uses a
 set of encoding files that used to ship with LM (lm-xx-os.enc).
 By way of a map file fragment, this allows to switch the fonts
 'en masse' from lining to oldstyle (and back) without the need for
 special metric files, which is a pretty nice feature.

Don't understand -- the vertical oldstyle digits, in general, have different
metric data (tabular digits differ only with respect to vertical dimensions).
So, the replacement of the encoding implies the replacement of TFMs. Or I
misundersood something.
 
 Those encoding files now apparently obsolete: they still refer to
 glyphs named e.g. `one.oldstyle', whereas in the current release of
 the LM fonts it really should be `one.taboldstyle'.

Yes, as I explained above.

 While writing a bug report, I noticed that these encoding files are no
 longer in the font distribution

No. We have a lot enough of TFMs and I'd be reluctant to add more. The only
resort I can see is to use OTFs...

 Is this trick to get oldstyle now officially unsupported?

Rather not, I've just learned about it. ;-)

 because there are no oldstyle-using TFM files in the LM
 distribution either ...

There are. As I mentioned, ts1 encoding contains oldstyle tabular figures.
Perhaps accessing them needs other tricks, but as long as 256-glyph fonts are
to be used, some glyphs must be accessed clumsily. e.g., proportional digits
(normal and oldstyle).

Cheers -- Jacko
___
If your question is of interest to others as well, please add an entry to the 
Wiki!

maillist : ntg-context@ntg.nl / http://www.ntg.nl/mailman/listinfo/ntg-context
webpage  : http://www.pragma-ade.nl / http://tex.aanhet.net
archive  : https://foundry.supelec.fr/projects/contextrev/
wiki : http://contextgarden.net
___


Re: [NTG-context] Latin Modern bug in lm-texnansi-os.enc

2007-05-24 Thread Taco Hoekwater


[EMAIL PROTECTED] wrote:
 Hi,
 
 On the ConTeXT mailing list, Idris posted some messages because he was
 trying to use oldstyle numerals in the latest Latin Modern release,
 and was experiencing two pretty big problems:

* really bad kerning around the oldstyle numerals
 
 We would be grateful if we could get more details (which font, which tfm,
 which encoding, pdftex or tex/dvips, on the screen or on the paper, etc.)
 My first guess is that proportional figures were used with tabular metrics...

Yes, but see explanation below.

 The implementation of 'automatic' oldstyle numbers in ConTeXt uses a
 set of encoding files that used to ship with LM (lm-xx-os.enc).
 By way of a map file fragment, this allows to switch the fonts
 'en masse' from lining to oldstyle (and back) without the need for
 special metric files, which is a pretty nice feature.
 
 Don't understand -- the vertical oldstyle digits, in general, have different
 metric data (tabular digits differ only with respect to vertical dimensions).
 So, the replacement of the encoding implies the replacement of TFMs. Or I
 misundersood something.

How it used to work:

   The tabular lining figures are replaced by oldstyle figures by
   means of an encoding file only. This worked fine (because the
   oldstyle figures were tabular as well).

The height and depth are normally not all that important to TeX,
and with the widths the same, there was no need for a different
metrics file, just a single encoding file is enough to give all the
LM fonts oldstyle (tabular) figures instead of lining figures.

This trick broke when the tabular oldstyle figures were renamed
and proportional oldstyle figures were added (to the non-typewriter
fonts. Both problems Idris had came directly from that the fact
that these changes make the encoding file invalid:

* the perceived kerning is wrong because the actual proportional glyphs
   have different sidebearings, and
* the monospace fonts gave an error because the glyphs were renamed,
   so the requested glyphs were no longer existent at all.

 While writing a bug report, I noticed that these encoding files are no
 longer in the font distribution
 
 No. We have a lot enough of TFMs and I'd be reluctant to add more. The only
 resort I can see is to use OTFs...
 
 Is this trick to get oldstyle now officially unsupported?
 
 Rather not, I've just learned about it. ;-)

Yes, I understand that now. Well, can you please consider starting to
support it then? ;-)

 because there are no oldstyle-using TFM files in the LM
 distribution either ...
 
 There are. As I mentioned, ts1 encoding contains oldstyle tabular figures.
 Perhaps accessing them needs other tricks, but as long as 256-glyph fonts are
 to be used, some glyphs must be accessed clumsily. e.g., proportional digits
 (normal and oldstyle).
 

I know there are way to many TFM  files already, I don't want you to
ship dozens (hundreds?) of new ones. But from a user's point of view,
having to switch to a different encoding for each number is not what
I consider 'supporting oldstyle digits'.

The trick I explained above works surprisingly well (even if the figures
are tabular, not proportional). If I had realized that those encoding
files were not created by you yourself, I would not have bothered you
at all, as it was not your problem to begin with. But that's life ...

Cheers, Taco


___
If your question is of interest to others as well, please add an entry to the 
Wiki!

maillist : ntg-context@ntg.nl / http://www.ntg.nl/mailman/listinfo/ntg-context
webpage  : http://www.pragma-ade.nl / http://tex.aanhet.net
archive  : https://foundry.supelec.fr/projects/contextrev/
wiki : http://contextgarden.net
___