[tex-live] cp8bit and old bugs

[Vladimir Volovich]

"PO" == Petr Olsak writes:

 PO> LaTeX users does not need to read the overfull \hbox messages (in
 PO> log and in terminal) in his own language? Texts in these messages
 PO> are not readable if ^^af sequences are used here.

 PO> If terminal is able to show 8-bit letters in choosen language
 PO> then it is more usable to configure TeX with 8-bit output.
 >> i'm not against allowing 8-bit output to terminal, log and files,
 >> because it is not contradicting the TeX specification by Knuth.
 >> 
 >> however, one should understand that such 8-bit text is not always
 >> useful, because of the difference in input encoding of the
 >> document and font encoding used (which is what the overfull \hbox
 >> messages appear in).

 PO> This is not exactly true. The overfull messages (and all others
 PO> messages) are converted by xchr vector (this vector can be set by
 PO> TCX or enctex).  It means: dvi encoding can be different from
 PO> terminal font encoding.  If the xord/xchr is set properly, you
 PO> can read overfull messages without problems.

TCX is global for processing of TeX document, but the document may
contain several different font encodings (e.g. T1 and T2A), and the
conversion by xord/xchr cannot achieve readability of overfull
messages in this case.

 PO> Because xord vector has to be set as inversion of xchr, you can
 PO> see that the reencoding only by inputenc at expand processor
 PO> level need not be the best choice.

but it's still better than changing xord/xchr since it relies only on
standard TeX features (thus is portable), and allows processing of
multilingual documents which is impossible with TCX. (also, TCX is
limited in that it only allows char to char translation, but sometimes
a character shall be expanded to some token sequence, e.g. some accent
command plus base character).

 >> in many cases this 8-bit text will look like unreadable 8-bit
 >> garbage, no more better than ^^af sequences.

 PO> No many cases, only rare cases... This problem can be occur in
 PO> 7bit TeX as well: \hbox to 0pt{text $\varrho$ text} prints the
 PO> percent character in overfull message. Important: this problem
 PO> does not interfere with legibility of the normal (non math) text.

here is a typical setup: i process LaTeX documents in koi8-r input
encoding, use the fonts in T2A font encoding, and have utf-8 terminal
encoding (i edit the documents in Emacs). it is impossible to make
the output to terminal readable in this case.

i even think that this is impossible with enctex :) because both input
and font encodings are single-byte encodings, and terminal encoding is
multibyte encoding.

Best,
v.