From ivan.castane-capel at nokia.com  Thu Sep 16 19:18:56 2010
From: ivan.castane-capel at nokia.com (ivan.castane-capel at nokia.com)
Date: Thu, 16 Sep 2010 19:18:56 +0200
Subject: [Qt-l10n] Length variants in qt
Message-ID: <B737AE8337D70F47B693AC85FA440B082A3B7B284A@NOK-EUMSG-02.mgdnok.nokia.com>

Hi all!

I'm part of a team in Nokia that is developing TDriver (previously MATTI), a GUI/API driven test automation tool for QT applications  and other platforms and we just hit some issues to do with localization and internationalization features of QT.

We are just extending our tool to allow for localized tests including the plurality and length-variant features. However we just noticed some unexpected behavior when using length variants and I could not find documentation for it in the qt literature.

To be more specific: we have a few users with localized applications that make use of the <lengthvariant> translation feature defined in the ts file dtd. We've seen that sometimes (or all, we just got some reports) the text attribute of widgets with more than one length variant contains all of the variants using the special char defined in the sources of linguist tools ("oe") as separator. I'm now trying to understand if that is the expected behavior and if so what should we change to keep our tool working.

Could anyone from this list shed some light to this issue?

Thanks in advance!
ivan

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.qt-project.org/pipermail/localization/attachments/20100916/c3578c83/attachment.html>

From oswald.buddenhagen at nokia.com  Thu Sep 16 19:46:16 2010
From: oswald.buddenhagen at nokia.com (Oswald Buddenhagen)
Date: Thu, 16 Sep 2010 19:46:16 +0200
Subject: [Qt-l10n] Length variants in qt
In-Reply-To: <B737AE8337D70F47B693AC85FA440B082A3B7B284A@NOK-EUMSG-02.mgdnok.nokia.com>
References: <B737AE8337D70F47B693AC85FA440B082A3B7B284A@NOK-EUMSG-02.mgdnok.nokia.com>
Message-ID: <20100916174616.GA24518@troll08.nokia.trolltech.de>

hi ivan,

On Thu, Sep 16, 2010 at 07:18:56PM +0200, Castane-Capel Ivan (Nokia-MS/London) wrote:
> We’ve seen that sometimes (or all, we just got some reports) the text
> attribute of widgets with more than one length variant contains all of
> the variants using the special char defined in the sources of linguist
> tools (“oe”) as separator.  I’m now trying to understand if that is
> the expected behavior and if so what should we change to keep our tool
> working.
>  
that's because the support for length variants is not implemented
pervasively (that's also why we didn't document it - it's sort of an
experiment and we are not really satisfied with it). the new declarative
stuff was supposed to consistently embrace the concept, but i have no
idea what became of it.
if you need to go more into depth, mail olivier goffart off-list (and
keep me cc'd).

regards


From ivan.castane-capel at nokia.com  Fri Sep 17 12:04:04 2010
From: ivan.castane-capel at nokia.com (ivan.castane-capel at nokia.com)
Date: Fri, 17 Sep 2010 12:04:04 +0200
Subject: [Qt-l10n] Length variants in qt
In-Reply-To: <20100916174616.GA24518@troll08.nokia.trolltech.de>
References: <B737AE8337D70F47B693AC85FA440B082A3B7B284A@NOK-EUMSG-02.mgdnok.nokia.com>
	<20100916174616.GA24518@troll08.nokia.trolltech.de>
Message-ID: <B737AE8337D70F47B693AC85FA440B082A3B7B2C42@NOK-EUMSG-02.mgdnok.nokia.com>

Hi Oswald,

I feared that was the case. Cool, as long as we know this we can now plan not to use it for now. I still had my doubts because the translation team at Nokia is generating ts files with locale variants, at least the Symbian team. I will let them know.

Thanks a lot for this info. Kind regards!
ivan


-----Original Message-----
From: Buddenhagen Oswald (Nokia-MS-Qt/Berlin) 
Sent: 16 September 2010 18:46
To: Castane-Capel Ivan (Nokia-MS/London)
Cc: qt-l10n at trolltech.com
Subject: Re: [Qt-l10n] Length variants in qt

hi ivan,

On Thu, Sep 16, 2010 at 07:18:56PM +0200, Castane-Capel Ivan (Nokia-MS/London) wrote:
> We’ve seen that sometimes (or all, we just got some reports) the text
> attribute of widgets with more than one length variant contains all of
> the variants using the special char defined in the sources of linguist
> tools (“oe”) as separator.  I’m now trying to understand if that is
> the expected behavior and if so what should we change to keep our tool
> working.
>  
that's because the support for length variants is not implemented
pervasively (that's also why we didn't document it - it's sort of an
experiment and we are not really satisfied with it). the new declarative
stuff was supposed to consistently embrace the concept, but i have no
idea what became of it.
if you need to go more into depth, mail olivier goffart off-list (and
keep me cc'd).

regards