texttops in CUPS 1.3

[Helge Blischke]

Opher Shachar wrote:
>>>>Opher Shachar wrote:
>>>>
>>>>
>>>>>In CUPS version 1.2.12 I can submit a print job with:
>>>>>
>>>>> lp -d MyPrinter -o document-format="text/plain;charset=iso-8859-8" an_iso-8859-8_file.text
>>>
>>>I set up a (virtual) Fedora 8 machine with CUPS 1.3.5
>>>    # yum list installed "cups*"
>>>    Installed Packages
>>>    cups.i386            1:1.3.5-2.fc8          installed
>>>    cups-libs.i386       1:1.3.5-2.fc8          installed
>>>    cups-lpd.i386        1:1.3.5-2.fc8          installed
>>>    cups-pdf.i386        2.4.6-6.fc8            installed
>>>
>>>and (sight) this all works well as in CUPS 1.2.12.
>>>
>>>If I may ask for anything, it is that you do not break this functionality at least for CUPS 1.3.x .
> 
> 
>>Helge Blischke wrote:
>>The "document-format=text/plain;charset=iso-8859-x" works in CUPS 1.3.5
>>if you have the respective charset file installed in (usually)
>>/usr/share/cups/charsets.
>>
>>But be aware that this might be changed in future releases.
> 
> 
> I hope it stayes on for all eternity :)
> 
> 
>>An alternative approach would be to install the texttops wrapper
>>from the cups links. You then need only specify the option
>>"encoding=iso-8859-x" and let CUPS do auto typing as usual.
> 
> 
> Not so fast. Hebrew has two flavors:
>   Visual (left-to-right) ordering aka iso-8859-8, and
>   Logical (right-to-left) ordering aka iso-8859-8-i
> Hebrew in UTF-8 is always (only) right-to-left ordering.
> Our applications print in left-to-right ordering (iso-8859-8) so the script needs to also do ltr --> rtl conversion.
> The documentation for texttops wrapper 1.01 has no mention of iso-8859-8-i and I suspect it does not do ltr --> rtl conversion (ie. treats iso-8859-8 as rtl).
> 
> Opher.

Like the author of Perl's Encode modules, I am not sufficiently aquainted to
the ltor and rtol conversion stuff. There is a Text::Bidi module by Moshe
Kamensky available on CPAN, but I've no experience with that.

By the way, the texttops filter implements some right to left handling
but AFAIR there is something more than just inverting the sequence
of characters.

Perhaps we could join our knowledge and hack a "well formed" solution?

Helge

-- 
Helge Blischke
Softwareentwicklung

H.Blischke at acm.org