Bug/regression with kashidas in 7.5.0.3 - but thread-starter no longer interested / therefore his "Please close this account"

12345678901234567890

Post deleted. I cannot find a way to close my account on this forum.

It’s great that people try newly released versions in the fresh branch (currently 7.5). Indeed, people who decide to become early adopters (as download page tells) would possibly see the regressions introduced in the new version, and file them as bug reports, because that’s naturally is what a user can easily contribute back to the project, and that is the only correct way to “have a fix”. Only thanks to these early adopters, later bugfix releases have a chance to become as robust and polished as possible.

And of course, until the problem is not fixed, the proper way is to use an older version. Again, after making sure that developers are notified (with test documents, screenshots, etc.) and thus have a chance to fix it.

@khaledhosny2 could possibly have an explanation what specifically has happened :wink:

Can you upload some example ODT, for example with one page of text?

12345678901234567890

12345678901234567890

A naive question:
Have you enabled Tools>Options, Language Settings>Languages Complex text layout? And of course set language to Arabic (national_variant) in your styles. Otherwise it is possible that Writer will not recognise Arabic and consider text as being in some Western script where there is no kasheeda.

12345678901234567890

I need a test document and the exact font. Most likely this is an AAT font not an OpenType one and HarfBuzz can’t tell us where it is safe to insert Kashidas, so we don’t insert any. Bug report is appreciated.

In < 7.5, we were inserting Kashidas when when it breaks text, now we only insert them when we are sure they won’t break text, and for AAT fonts we don’t know whether it is safe to insert Kashida or not so we don’t insert any (AAT fonts are rare, basically only the few old macOS fonts that Apple still ships, and may be a handful third party fonts).

12345678901234567890

I installed the font XB Kayhan and then open your document and it seems OK. The same when I changed the font to Scheherazade or Scheherazade New.
But for KFGQPC Uthman Taha Naskh, Amiri and Noto Naskh Arabic it showed “kashidas”, but always for different words. I’m not sure, but maybe it is related to the ligatures in the font.


Win10x64 Libre 7.5.0.3

Like I suspected. Please report an issue on bugzilla. On the meantime, I suggest using different fonts. XB fonts are probably the only third-party Arabic AAT fonts.

12345678901234567890

“If” your documents are based on the same template (so that the styles confer a uniform layout to all of them), all you need to do is to modify one or a few styles in the template. The documents are automatically updated next time you open them.

When you maintain a collection of several hundreds document, a strict methodology is mandatory. Not using template is equivalent to direct formatting text but on a much larger scale.

Even if it is tedious and boring, it might be time to adopt a “professional” approach to document management.

12345678901234567890

Which is the wrong way of using it.

Are you sure this can’t be solved with a judicious collection of styles. There are surely not several hundreds different layouts. If they number less than ten, it can be handled elegantly with styles. Not all people will use all styles but they’ll find what they need in the collection.

12345678901234567890

Omg. Yes, reading is hard.

You had been asked to file a bug. Twice. By two (!) actual developers who happened to read this: myself and @khaledhosny2 (who is the expert in fonts and layout and CTL (using CTL, and being one of principal contributors both to LibreOffice and to Harfbuzz). Asking for a bug means we agree that the issue exists. Khaled also suggested you a workaround in the meanwhile, knowing the root cause, and in the impression that people who ask here usually need some solutions that would work now. All that you replied was rudeness.

12345678901234567890

I read it, and I didn’t see why anyone would need to wait, but it’s your decision. It just feels sick when someone asking for help gets a suggestion to file a bug report (with a note that the version is new an regressions are expected), then tells that they will wait, then gets another reply describing the problem in depth (which mentioned that the problem is specific to very few fonts), and then posts a series of comments blaming people trying to help.

12345678901234567890