[Issue] Combining diacritics causes letter duplication

[dm_dobrowolski]

There seems to be a bug in OOo 2.4.1 Writer: inserting a combining diacritical mark over a symbol causes the latter to be duplicated with a slight shift rightwards and an unexpected blank place after. The diacritics appears over the second copy. A screenshot with a russian word Проба ("Test") and macron over "cyrillic o" letter attached below.

The problem appears not always, but often, especially when trying to type a longer text (or when Jupiter is in Gemini, or when they sing Halleluyah in the nearby church, or... - I really don't understand the logics ). Some folks say it has some relation to proceeding OpenType tables

The problem was first described here:
viewtopic.php?f=7&t=1502&p=6087
(if you read in russian you may look here too:
http://community.livejournal.com/old_ru ... 6#t1353896)

There was a correspondind issue:
http://www.openoffice.org/issues/show_bug.cgi?id=81892
But this issue was closed on Tue Sep 25 2007, by the one, who concerned that "It works fine in OOo current", which is not quite right, as you can see.

This is not quite critical, but really disturbing, especially if you're a philologist dealing with slavic languages, and you do need diacritics over cyrillic letters, which are not represented in Unicode.

So (finally!) the question comes: can I reopen an issue, or should I start the new one once again?

Thanks,

[Hagar de l'Est]

When you post a comment in an issue, there is usually an option to reopen it when it has been closed. But there is no such option in that one, don't know why, perhaps because it has been closed WORSKFORME.

So I think you should open a new one, mentioning the Issue 81892 (NB: the "Issue #####" string is automatically recognized and will be changed to an hyperlink to that issue).

NB: I move your post in the Writer forum, feel free to change your thread title so that other users with such an issue can spot it.

[acknak]

I would not start a new issue if you're sure that it's the same problem.

Instead, just add a comment to the existing issue asking that it be reopened and explain why you feel it was closed improperly. Be sure to include a test document and/or screenshots demonstrating the problem.

It may be that issues can only be re-opened by a developer because too many users were changing the status improperly. I've seen it happen many times.

[dm_dobrowolski]

feel free to change your thread title so that other users with such an issue can spot it.

...just add a comment to the existing issue asking that it be reopened and explain why you feel it was closed improperly

Thanks. I've renamed the thread and added the comment in the issue. Now waiting...

[push]

As I understand, the topic was closed because the issue is resolved in OOo 3.0 which is not yet a public release. (I haven't tried it myself, though).
So indeed, it is not resolved in OOo 2.4 but I'm not sure it should be reopened.

[acknak]

Hmm, yes, it seems this is a little more subtle.

First, the screenshot is nice, but we really need a document, or at least a description of what characters you used in the document. Otherwise, there's no way anyone else can try to reproduce the problem.

Second, this comment (which I skimmed over at first) may be a key here (my emphasis):

This is... disturbing, ... [if you] need diacritics over cyrillic letters, which are not represented in Unicode.

I'm not a font expert at all, but I don't think you're ever going to see OOo support combining characters outside of a Unicode context. If a font supports the combined characters as a separate glyph, sure, but I think Unicode provides a standard means to handle character combining properties. Why should OOo support more than one means to do the combining?

Perhaps I misunderstood your statement.

[Bhikkhu Pesala]

I was able to reproduce the problem easily when it was reported in this earlier thread.

Combining diacritical marks are part of the Unicode system, so Oo should support them.

This is not just a cosmetic problem since PDF output shows the same problem as is seen in the document.

My recommendation would be to design and use custom fonts for this kind of work. The Charis SIL fonts may provide what you need. They can do some very intelligent positioning of diacritics. I still see the same problem, but there are ways to work-around it. What seems to work is to reapply the font with direct formatting after typing the text.

The attachment shows how Charis SIL can intelligently position the diacritics over the h, or combine more than one diacritic. It will also centre them over characters of different width, and reduce the length of the macron when combining it with narrow letters like i. Garava is unable to centre the macron properly over the narrower letter r, or raise it for the h, though it fits OK on the v.

CharisSIL Diacritics.png (3.39 KiB) Viewed 323 times

[acknak]

Combining diacritical marks are part of the Unicode system, so Oo should support them.

This is not just a cosmetic problem since PDF output show the same problem as is seen in the document.

Both editing and PDF export work fine for me.

cyrillic_combining.png (18.18 KiB) Viewed 321 times

That is just a sequence of Unicode characters:

U+043E
U+043E U+0304 (COMBINING MACRON)
U+0436 U+032F (COMBINING INVERTED BREVE BELOW)
U+043B U+0308 (COMBINING DIAERESIS) U+032B (COMBINING INVERTED DOUBLE ARCH BELOW)

First in the default font (TNR), then copied and changed to font "SIL Doulos IPA".

[dm_dobrowolski]

First in the default font (TNR), then copied and changed to font "SIL Doulos IPA".

Quite right, It works at my PC too, but SIL Doulos IPA is an 8-bit font with insufficient repertoire of symbols, and it won't be maintained by SIL team anymore, so needed symbols won't ever appear. And if you try a 16-bit Unicode font, you'll see the effect.

First, the screenshot is nice, but we really need a document

Sorry for not giving you the files

diacr_bug_LL.odt

The bug appears rather in a long texts, so I've taken a part of my last published article, and added two diacritics (U+0300 & U+0301) to arbitrary words in the fisrt and in the last paragraph. Font used - Linux Libertine.

(15.02 KiB) Downloaded 5 times

diacr_bug_LL.pdf

Previous file PDF-ed. "Buggy" places highlited with yellow.

(32.27 KiB) Downloaded 2 times

diacr_bug_OS.odt

The same text in Old Standard TT font, used to produce the screenshot.

(20.58 KiB) Downloaded 3 times

Paths for the fonts used:
Linux Libertine = http://sourceforge.net/projects/linuxlibertine
Old Standard = http://www.thessalonica.org.ru/en/fonts.html

[acknak]

Your document looks good for me, as does the PDF I exported from it.

The text layout is slightly different for me; I'm not sure why.

Edit: Oops, it's different because I don't have the language support for OOo to do the automatic hyphenation!

diacr_bug_LL_f9.png (39.19 KiB) Viewed 322 times

[dm_dobrowolski]

So, the first result: the bug exists in Windows version of OOo only (or is caused by Windows font / graphics mechanics).

Thank you .