msevior ([info]msevior) wrote,
@ 2007-01-15 07:36:00
Previous Entry  Add to memories!  Tell a Friend!  Next Entry
AbiWord-2.5.0 complex script testers please?
As noted by Marc "uwog" Maurer and hub AbiWord-2.5.0 was released today.

Like every free software project, we really appreciate testing and feedback from our users. However for the 2.5.0 release cycle we have a special plea to speakers of Arabic, Farsi, Hebrew, Indic, Thai and other complex script languages. One of the really big efforts in the 2.5.0 cycle is to move to a new pango-based text renderer for Linux. This is primarily the work of Tomas Frydrych,, ex-biblical scholar, now OpenedHand developer. With this we can in principle fully support all the languages supported by pango. However none of us core Abi-Hackers has much knowledge of these languages so we really need help from native speakers to provide feedback and to let us know what doesn't work.

Your help is much appreciated.

(Post a new comment)


[info]kerrick
2007-01-15 07:30 am UTC (link)
I compiled with --enable-debug and I get an assert:

**** (5) Assert ****
**** (5) iGlyphsStart >= 0 && iGlyphsEnd >= 0 at gr_UnixPangoGraphics.cpp:1176 ****
**** (5) Continue ? (y/n) [y] : y

When selecting a letter with a vowel. (Use keyboard layout "Israel Lyx" and type "n" "shift-n" for example or paste the following מֳ for example)

It does render so much better than 2.4 though. The vowels all go in their correct places. Kudos to Tomas.

(Reply to this) (Thread)

(Anonymous)
2007-01-15 11:07 am UTC (link)
@kerrick; could you file suchs things in our bugzilla at bugzilla.abisource.com ? Thanks a lot for building it with debug info as well :)

(Reply to this) (Parent)
Nothing new for Japanese?
(Anonymous)
2007-01-15 11:37 am UTC (link)
I guess there's been no improvement for Japanese? I just tried the devel release and it's just the same as always - doesn't switch to a typeface that has the glyphs, typing and conversion doesn't happen inline like it should but in a tiny window somewhere below the real window and so on. Or have I just missed a build option?

(Reply to this) (Thread)
Re: Nothing new for Japanese?
[info]msevior
2007-01-15 12:19 pm UTC (link)
Unfortunately not. It sounds like Japanese is rendered correctly but that the Input Method leaves a lot to be desired.

Can you give an example of a Gtk+ program that does the right thing for Japanese text input?

Thanks for your feedback!

(Reply to this) (Parent)(Thread)
Re: Nothing new for Japanese?
(Anonymous)
2007-01-15 02:05 pm UTC (link)
Actually, I believe abiword is the only gtk-based program I've encountered which doesn't do the right thing. I had hopes that with a switch to Pango abiword would resolve these issues as well; it's a bit more complicated, I guess :) Just to give you a couple of names, Gnome-terminal does the right thing, as does Gedit. For non-gnome apps, Firefox, gvim and granule all work fine. OOwriter does this correctly, but has other issues (the settings UI is cryptic and it gets the relative font sizes wrong with kana being taller than roman).

One quirk is of course specifically with the replacement font. For all of the above (and 99% of all apps, period) they just need to switch to what the font config system tells them (and I have no idea how it determines it in turn), but with a word processor it may become a bit more complicated. The user may want to be able to set the preferred Japanese font more or less independently from the roman font. If you do that, don't look to OOwriter for UI giudelines; it's a mess, where just figuring out what setting changes what feature is fraught with frustration.

For me, I'd ideally want see the font selector item switch to the currently set Japanese font when I activate Japanese input method. If I change it, I change the preference - and only for the Japanese font. When I switch back, I get the roman setting again.

Happy to give more feedback if it can help!

BTW, on a far off wishlist is furigana (ruby text, set over or under each character for pronunciation indication) and ability to write vertically from right to left. Oh, and a pony :)

(Reply to this) (Parent)(Thread)
Re: Nothing new for Japanese?
[info]msevior
2007-01-15 02:29 pm UTC (link)
Thanks very much for this detailed information.

The problem we have is that none of the core of the Abiword developers actually speak or write Japanese so we don;t really know what the correct thing is.

The pointers to the gtk-apps are potentially useful because we can look at the source code and see how they do their input methods plus ask questions of the developers.

I invite you to open a bug in our bugzilla, "Japanese Input Method not adequate for Unix", and copy the text you've written to it. It will be a good way to track progress and maybe even fix it.

http://bugzilla.abisource.com/

(Reply to this) (Parent)(Thread)
Re: Nothing new for Japanese?
(Anonymous)
2007-01-15 03:15 pm UTC (link)
Sorry - I truly thought I'd entered a bug about this a long time ago, but of course I can find no trace of it now.

Here's the bug:
http://bugzilla.abisource.com/show_bug.cgi?id=10756

(Reply to this) (Parent)(Thread)
Re: Nothing new for Japanese?
[info]7wrc
2007-05-02 09:38 pm UTC (link)
Thanks for details.

(Reply to this) (Parent)
Hi :)
(Anonymous)
2007-01-15 01:22 pm UTC (link)
Hi,

I would love to test AbiWord for indic scripts, specially the ones which are based on Devnagari script.

Do you have any test plans already set up? Or should I propose some?

Letme know if its okay...

Regards,
makuchaku (mayank.gnu at gmail dot com)
http://www.makuchaku.info/blog

(Reply to this)

Create an Account
Forgot your login?
Login w/ OpenID
English • Español • Deutsch • Русский…