Hebrew Diacritics

[Yaron]

Hello, recently I gor some reports about misaligned diacritics in Hebrew,
I was unable to generate a screenshot but I have seen this happen.

This bug has several aspects:
1. I cannot enter the diacritics in the accepted ways, either by having Caps Lock selected and then press Shift+numbers or -=\.
2. I cannot use their key combination (Alt+0206 which is Maqaf for example).

So I wrote the text in an external editor (Notepad) and copied and pasted the text in
The diacritics were all misaligned.

Please notice, this is a Windows specific bug, in Linux it works perfectly while using the LyX keyboard (Inserting diacritics with this layout only requires pressing Shift with one of the letters, this layout does not have English Capital letters support at all, you need to switch layout in order to type in English)
Meaning that there's no need to move to the 3rd selection mode in order to insert diacritics.

I would love it if someone could please confirm this bug under Windows, these errors are possibly Arabic related as well.

I will try and provide screenshots as fast as I can.

In the meanwhile I can provide a Hebrew word with diacritics and an Arab word with diacritics:

Hebrew - בְּרֵאשִׁית, בָּרָא אֱלֹהִים, אֵת הַשָּׁמַיִם, וְאֵת הָאָרֶץ (Bereshit bara Elohim et hashamayim ve'et ha'arets - In the beginning God created heaven and earth) (Holly Bible/New Testament)

Arabic - بِسْمِ ٱللهِ ٱلرَّحْمٰنِ الرَّحِيمِ (b-ismi-llāhi r-raḥmāni r-raḥīm - In the name of God, Most Gracious, Most Merciful) (Holly Qur'an)

Kind regards,
Yaron Shahrabani

[jazzynico (jazzynico)]

Misalignment confirmed on Windows XP, Inkscape 0.47 and bzr rev. 9369.
Reproduced with Arial, TNR and Lucida Unicode.
Not reproduced with OpenOffice, on the same computer.

[su_v (suv-lp)]

confirmed with Inkscape 0.47+devel r9388 on OS X 10.5.8
reproduced with Arial Unicode MS and Corsiva Hebrew

[David Mathog (mathog)]

There are still issues with the rendering of Hebrew with diacritics on both WIndows XP and Linux. It seems to be related to the font chosen, at least in current trunk.

Load the varady_hebrew_with_vowels.svg (an example file from Aharon Varady's SVG test site), select all of the characters
in the one text object, then change fonts.

Fonts that do not work:
Arial, Courier New, Liberation Mono, Liberation Sans

Fonts that do work:
Ezra Sil SR (or whatever it fails over to, since that font is not actually on my systems), Bitstream Charter,
Century Schoolbook, Verdana, Trebuchet MS

In all cases when a font "does not work" the problem is that the diacritic acts like a normal glyph, with a
width and advance. So the vowel ends up under a "ghost" letter, rather than the intended one. This also happens
when the ^U method is used to enter one of these symbols. For instance, ^U5b8 does this.

On a somewhat related note, there seems to be no convenient way to edit these sorts of diacritics. They may be entered using the the ^U mechanism, but they may not selected separately from the base letter, so the only way I have found to change one is to delete the base, which takes the diacritic with it, and then reenter the base, and then the change. This is true even for the fonts that do not work, where the diacritic acts in other ways (spacing) like a separate character.

[David Mathog (mathog)]

I think I can explain why this is happening, but I do not know where the problem is. The glyphs in question fall into the Unicode category"Mn" == "Mark, nonspacing". This attribute may or may not be encoded in the font somewhere. Most often I think, not. Even if it is, Freetype may not know how to extract it. As a result, either Inkscape, Pango, or Cairo is not doing the right thing when it attempts to display these glyphs. Presumably this also affects some or all of the 1000'ish other Mn unicode glyphs used in other languages.

The complete list of such glyphs may be found with this command:

  wget -q -O - http://www.unicode.org/Public/6.2.0/ucd/UnicodeData.txt \
    | grep ';Mn;'

There is an Mn table already present in Inkscape in the src/dom/ucd.cpp file. Either something is not using it or not using it properly.

Since essentially every font I tried was broken for Hebrew in one way or the other in the end I had to load Ezra SIL SR (google
for it, freely available) onto my systems. (Google for it, free font.)

[Beluga (buovjaga)]

Tried changing to Liberation Sans in varady_hebrew_with_vowels.svg, but I am not seeing the ghost letters. Please re-test.

Arch Linux 64-bit, KDE Plasma 5
Inkscape 0.92+devel 15099 (GTK3)

[jazzynico (jazzynico)]

Not reproduced on Windows 7, Inkscape 0.47, 0.48.5, 0.91 and 0.92.
Fixed in the system fonts?...

[David Mathog (mathog)]

Font support for Hebrew is extremely variable. At the good end you have Ezra SIL SR, which works perfectly (see the attachment). Liberation sans is slightly broken (not shown), with just one vowel slightly offset. Arial is really bad (also in attachment).