Arabic fonts should respect CGJ about arabic

[behdad]

/cc @roozbehp

[kmansourMT]

I think this bug report is asking for a way to inhibit certain special combinations of marks through the use of an intervening CGJ mark (034F). A common example could be the combination of Shadda (0651), an upper mark, with Kasra (0650), a lower mark. Following is the typical combined mark of Shadda-Kasra:

The string for the above graphic is 628 25cc 651 25cc 650 3a 20 628 651 0650 3a 628 25cc 0650 25cc 651 3a 20 628 0650 651.

Now, if we were to insert a CGJ mark between the two marks, Shadda and Kasra, we could keep the combination from forming as in:

Under bug 85873, the above sample character string (628 25cc 651 25cc 064e 3a 20 628 651 34f 064e 3a 628 25cc 064e 25cc 651 3a 20 628 064e 034f 651) uses two upper combining marks (Shadda 0651, Fatha 064E). Its rendering above shows that CGJ not only inhibits the formation of the combination, but also has the unintended effect of interfering with the mark-to-mark positioning between the Shadda and the Fatha.

[behdad]

@kmansourMT What system did you use to generate the images?

[kmansourMT]

Monotype's OpenType interpreter was used.

[ALloydMT]

We added u+034F as a non-spacing mark.

Without u+034F, using the following code 0627 0645 0651 064D gives:

Including u+034F in the following code 0627 0645 0651 034F 064D we get:

[behdad]

Thanks @ALloydMT. Can you also attach results for:

0627 0645 064B 0651
0627 0645 0651 064B
0627 0645 0651 034F 064B
0627 0645 064B 034F 0651

[ALloydMT]

Hi Behdad, here are those combinations.

0627 0645 064B 0651

0627 0645 064B 034F 0651

0627 0645 0651 064B

0627 0645 0651 034F 064B

[simoncozens]

This appears to be done.