[XeTeX] [Fwd: Re: FreeSerif not working for me in Devanagari]

[Steve White]

I finally got Neal's examples working, too.  (The problem was, I had
altered the header tex file he provided, so that it would run with my
set of fonts, but had unwittingly also cut out one instance of
RomDev).  The process is: obtain the RomDev.map file, compile it with
'teckit_compile -u RomDev.map', place the resulting RomDev.tec file in
the same directory as the .tex files.  The xelatex the main tex file.

The result is pretty to my eyes; the conjuncts are working, as near as
I can tell.

Concerning Zdenek's comments:

The Devanagari feature lookup tables in FreeSerif are arranged so that
modern forms are default, and the more extensive Sanskrit forms must
be turned on by explicitly specifying Sanskrit as the text language.
(It is safest always to specify the language.  Unfortunately, xelatex
complains very confusingly if no tables are explicitly turned on by
the specified language.)

FreeSans is indeed derived from Gargi, but it has been extensively
modified.  It (is intended to) have complete support for Hindi and
Sanskrit, as well as Marathi.  If any conjuncts in FreeSerif are
missing in FreeSans, it is a bug.

Gargi has undergone big improvements this year, thanks to Monika Shah.
I have seen some interim versions, and have been told it will be released soon.
(The glyphs in FreeSans are drawn from earlier versions of Gargi,
and the font has developed independently since.)
My understanding however is that Gargi was originally developed by
Hindi speakers.

Cheers!


On Sun, Sep 9, 2012 at 6:35 AM, Neal Delmonico <ndelmonico at sbcglobal.net> wrote:
> I am not sure this made it to the full list.  Therefore, I am forwarding
> it.
>
> My problem with FreeSerif has been solved.  Everything is working now,
> even the page headings with the help of FakeSlant.  I think it is the
> best font yet for those of us working with Devanagari on a regular
> basis.
>
> Best
>
> Neal
>
>
> ---------- Forwarded message ----------
> From: Zdenek Wagner <zdenek.wagner at gmail.com>
> To: Steve White <stevan.white at googlemail.com>
> Cc: <ndelmonico at sbcglobal.net>
> Date: Sun, 9 Sep 2012 01:05:01 +0200
> Subject: Re: [XeTeX] FreeSerif not working for me in Devanagari
> 2012/9/8 Steve White <stevan.white at googlemail.com>:
>> Hi Neal,
>>
>> I'm very pleased to hear it's working for you!
>>
>> Could you please write to the mailing list, to let them know?  (To
>> date, the advice has been "don't use FreeFont".)  It would be great to
>> see your working examples there, too.
>>
> I had some communication with Neal off list so I will summarize now.
>
> 1. Devanagari is a script used for several languages. They differ
> mainly in the repertoire of ligatures used, the full list being used
> in Sanskrit. Up to now, there anre Snaskrit fonts (not usable for
> Hindi, Marathi etc. because the users without education in Sanskrit
> will be unable to read them), Hindi fonts (not usable for Sanskrit due
> to missing ligatures), Marathi fonts (not usable even for Hindi), yet
> all of them use the Devanagari script. FreeSerif is the "new
> generation" font because it supplies the Devanagari script and the set
> of ligatures are set according to the language. FreeSans does not
> contain all Sanskrit ligatures.
>
> 2. The Devanagari block is missing in the italic shapes in both
> FreeSerif and FreeSans. In order to use them, AutoFakeSlant has to be
> specified. The default value is 0.2 if not given.
>
> Written shortly, Neal Delmonico uses Charis SIL as the default font
> and FreeSerif as the Sanskrit font. In order to have it work with all
> Sanskrit ligatures and italic, the header shouldbe as follows:
>
> \usepackage{xltxtra}
> \usepackage{polyglossia}
> \setmainfont{Charis SIL}
> \newfontfamily\sanskritfont[Script=Devanagari,Mapping=RomDev,Language=Sanskrit,AutoFakeSlant=0.195]{FreeSerif}
>
> Mapping=RomDev is only needed if the text is input in transliteration,
> if the text is typed directly in UTF-8, it is not needed.
>
> For those who need more information:
> FreeSerif contains the Velthuis glyphs (I hope that the PFB files
> hand-tuned by Karel Piska were used). Positions of matras were
> precisely adjusted. Some characters with nuktas require different
> positions of u and uu matras in order to be readable and the nukta
> were visible and Steve made it. Consonants ka and pha need different
> position of e and ai matras. All this is done.
>
> FreeSans is derived from Gargi. It seems to me that Gargi was a
> Marathi font because unlike Hindi, Marathi does not use characters
> with nuktas. The characters with nuktas were present but there half
> forms were missing. Thus the font was almost unusable fot Hindi. I
> know that nuktas are often omitted in Hindi, ja is often used instead
> of za (some people even pronounce jaruur instead of zaruur), pha is
> often used instead of fa, qa is almost always replaced with ka. The
> correct half forms were added to FreeSans. Nowadays fonts often omit
> the classical kra ligatures. I have asked my Indian friends and they
> replied that the they would prefare the classical shape. It was
> therefore made by Steve. And as with FreeSerif, positions of matras
> were precisely tuned.
>
> I forgot to mention positions of anusvaras. They are used in Hindi in
> the oblique case in plural and with verbs in plural feminine. The
> situation may be quite complex with some words and the position of
> anusvaras in all possible cases was tuned in order to make the words
> readable.
>
> Indian typographers use larger spacing preceding punctuation. In
> Sanskrit only dandas and double dandas are used but nowaday's
> languages accept also question and exclamation marks. The same spacing
> is expected (a few years ago I read an article written by an Indian
> typographer but I am not able to find it now). Majority of Devanagari
> fonts do not take it into account. Native users thus tend to enter a
> space preceding punctiatiom marks which leads to incorrect line breaks
> where punctuation may appear at the beginning of a line. It would be
> necessary to enter a fixed-width nonbreakable space but it is not
> usually available on a keyboard. FreeFont uses correct spacing. The
> same punctuation marks behave properly both in the Latin and
> Devanagari scripts (both FreeSans and FreeSerif, tested in a longer
> Hindi text).
>
> It was a huge amount of work but now all aspect of typesetting in
> Devanagari are properly set. Due to Steve's big care it is now the
> most beautiful Devanagari font.
>
> Remember that Unix distributions often come with an older version of
> GNU FreeFont. If you want to use the current version as distributed
> with TeX Live 2012, you have to delete the system fonts and follow the
> post-install actions as given in the TeX Live manual. The new version
> of the GNU FreeFont will then be available to the system, no
> applications relying on the existence of the font will be broken
> (verified by me in four different Linux distributions).
>