[tex-live] [Fwd: [Fwd: ec.enc, cork.enc and tcaron]]

[Vit Zyka]

Hellow Taco and others, please read bellow:

Taco Hoekwater wrote:
> 
> Hi Vit,
> 
> Jim Hefferon wrote:
>> Please forgive the forward; I don't understand anything about encodings
>> myself so I can only pass on the report whole.  Vit Zyka wrote me the
>> message below. 
>> -------- Forwarded Message --------
>> From: Vit Zyka <vit.zyka  munging  gmail.com>
>> To: ftpmaint  munging  alan.smcvt.edu
>> Subject: [Fwd: ec.enc, cork.enc and tcaron]
>> Date: Mon, 14 Apr 2008 13:12:34 +0200
>> Dear Jim,
>>
>> I do not know who is responsible for a next bug; in the cork.enc file 
>> I find address <tex-fonts at tug.org> but my mail returned back 
>> undelivered. So I am sending this bug to you are the CTAN maintainer. 
>> Please, might you correct it yourself or resent to a responsible person?
>>
>> There is wrongly named the glyph 'tcaron' by 'tquoteright' in the files
>> ec.enc and cork.enc. It is really bug, see e.g.
>> http://www.fileformat.info/info/unicode/char/0165/index.htm
>> Due to the fact the glyph is simply missing using some fonts in the
>> resulted documents.
> 
> ... and may be present in others.
> 
> Encoding files that are syntactically correct and succeed in creating
> a PostScript array of length 256 containing names are bugfree when they
> are used with a font that uses these names to indentify charstrings.

Yes, Taco, you are right the bug in [ec|cork].enc sould be solved in 
conjunction with the fonts.

Let as look to the TeX Live 2007 temxmf-dist/fonts directory. Command
   grep -S "tquoteright" *.*
responds only
.\afm\ibm\courier\cour.afm:C -1 ; WX 600 ; N tquoteright ; B 73 -14 646 
563 ;
... next 7 font variant
.\afm\ibm\times\nntb8a.afm:C -1 ; WX 521 ; N tquoteright ; B 0 -6 493 677 ;
... next 7 font variant
.\enc\dvips\base\cork.enc:  /tquoteright /tcedilla /uhungarumlaut /uring
.\enc\dvips\base\ec.enc:  /tquoteright /tcedilla /uhungarumlaut /uring
.\enc\dvips\base\tex256.enc:/tcaron %   180 % /tquoteright

So:
1. tex256.enc has been corrected yet.
2. The only fonts that use buggy name are ibm/courier and ibm/times
    Total number of afm files that use tquoteright is 16.
3. Total number of afm files that use tcaron is 12848(!).
4. cork.enc is not used in *.map at all
5. ec.enc is used for 611 fonts in *.map
6. none of ibm fonts in tfm/ibm/*
    (cour,courb,courbi,couri,cr-pc8,crb-pc8,crbi-pc8,cri-pc8,
     nntb8a,nntbi8a,nntr8a,nntri8a,ntnb8a,ntnbi8a,ntnr8a,ntnri8a,
     and their variants)
    is present in *.map; so no change for these fonts is needed if
    we correct the enc files.
7. Except of missing of \v t in the text, Zdenek Wagner summarized the 
next problems with 'tquoteright':
 > * if afm2tfm or a similar program is asked to generate
 >   small caps font by scaling the uppercase characters.
 >   When creating small caps /tcaron the program will
 >   look for /Tcaron and scale it.
 > * The character has meaning of /tcaron, its Unicode name is
 >   LATIN SMALL LETTER T WITH CARON.
 > * when searching PDF or when cutting&pasting the texts
 >   from PDF via a clipboard. If /toUnicode map
 >   is not inserted, the algorithm relies upon the postscript
 >   names and /tcaron is expected here.

I propose:
A) rename (for hypothetic use with font like ibm)
      en.enc -> ec-old.enc
      cork.enc -> cork-old.enc
B) correct glyph name /tquoteright -> /tcaron in
      ec.enc
      cork.enc

These steps should not cause any back compatibility problems. Despite 
this if some occurs we will solve it by using *-old.enc in map files or 
discussing the glyph name with the font authors.

Alternatively, we can create new file ec-correct.enc and to change six 
hundreds lines or so in *.map to point to this new file. I see these 
drawbacks:
a) more work, more affected files
b) confusion for users since there exist manuals that advise to use 
ec.enc file (e.g. afm2tfm, pdftex).

Preserving on the present encoding vectors is a bug that causes problems 
to users. Please, let us try to solve it.

Regards
Vit Zyka

> Whether or not these particluar names are used in any font is another
> matter, and that is why the latin modern fonts come with their own
> files (lm-ec.enc, I believe). Various fonts use various 'names' for
> their glyphs, and an .enc file *has to* abide to whatever convention
> the font designer has chosen.
> 
> This cork.enc may show buggy behaviour when combined with the font
> it is tied to in your psfonts.map or pdftex.map (and if that came from
> texlive, then there is a bug in the map file) but the encoding file
> itself is still correct even in that case.
> 
> Hope this helps,
> 
> Taco
> 
> 

-- 
=========================================================
Ing. Vít Zýka, Ph.D.           TYPOkvítek, Czech Republic

computer vision application  aplikace pocitacoveho videni
database publishing                databazove publikovani
scientific book typesetting     sazba odbornych publikaci

tel.: (+420) 777 198 189       www: http://typokvitek.com
=========================================================