Please provide links to:
- the font you use (Sanskrit2003.ttf)
  or are you using the font from www.omkarananda-ashram.org?
 
- expected results (using another program if required)
  as images

- actual result as images

1. The font can be downloaded from 
http://www.omkarananda-ashram.org/Sanskrit/itranslator2003.htm#dls

2. The expected result may be generated by any browser which can deal with the Unicode sequence I gave: क्&2340;  What you will see (if an appropriate font has been installed) is a single glyph which is a conjunct of the glyph for "k" (&#2325) and "t" (&#2340).

3. The different results (text from the browser or image) may be seen on my testpage
http://leuchtraum.de/translation/test/devatest.php?text=kt
Please see also the source of the page. 
Try also input of "krtsna" or other conjuncts. (This test page is not perfect but hopefully sufficient to show the problem).

4. The Unicode directives for Devanagari are given in http://www.unicode.org/versions/Unicode4.0.0/ch09.pdf
Table 9.2 on p. 229 is of relevance; also the pages before.

Thanks for the links, they are mostly what I'm using to port the Freetype backend to the newly added Unicode support in HEAD (php6).

Did that ever work? 

For what I see, it seems like I will be able to really fixed that only in php6 using unicode (unicode strings passed to FT with correct glyphs set).

Did that ever work? take #2

It's a new project, so I do not know whether that ever worked.

Is this problem solved? I tried using several Bengali unicode fonts. The problem exists.

This issue appears to have been fixed as of PHP 5.3.0, see
<http://imgur.com/JT577Q7,yKzJI0I>, or do I misunderstand the
problem?

No feedback was provided. The bug is being suspended because
we assume that you are no longer experiencing the problem.
If this is not the case and you are able to provide the
information that was requested earlier, please do so and
change the status of the bug back to "Re-Opened". Thank you.

Problem still remains. Image shown on imagur is incorrect.

Please refer http://unicode.org/faq/indic.html#17

This will provide you enough information on how 'virama' combinations should be rendered. 

I could provide any supporting information you may need.

A direct browser based text to image:

http://utils.net/caption/fidd-utf8:hi_in/fid-r/cn-1/fidd-utf8:hi_in/t-AnyHindiUnicodeStringHere

On every reload, it will use a random hindi unicode capable code.

For any other language change utf8:hi_in to utf8:<2letter_language_code>_<2_letter_country_code>

Thanks, now I got it. The problem isn't particularly related to
Bengali, but rather to combining characters in general, which are
not supported by imagettftext().

I think, you got it. Problem is with most Indic and Arabic languages.

Please note that most of these fonts carry GPOS and GSUB tables. Font tables also have information related to substitution of glyph combinations with a different glyph.

Refer: http://fontforge.github.io/gposgsub.html

we are running into this bug when writing
क्षत्रिय on an image using imagettftext
imagettftext can not write क्ष

I created a gist which clearly shows the misbehavior[1].  I'm
pretty sure this is actually an upstream issue, so should be
reported there[2].

[1] <https://gist.github.com/cmb69/0355b94c13631b7b1a64df21388039d5>
[2] <https://github.com/libgd/libgd/issues>

I have forwarded this issue to libgd[1], and suspend this ticket
for the time being.

[1] <https://github.com/libgd/libgd/issues/758>

Closing as duplicate of <https://github.com/php/php-src/issues/17352>.