Apache OpenOffice (AOO) Bugzilla – Issue 20537
Move strings from scp to officecfg for better language pack support
Last modified: 2004-07-14 12:45:46 UTC
In language pack scenarios (e.g. Ximian, Debian, ... packages) the menus for New and Wizard are not correctly translated. http://installation.openoffice.org/servlets/ReadMsg?msgId=627719&listName=dev outlines possible solutions. A patch that adds the strings to officecfg is at http://ooo.ximian.com/patches/OOO_1_1/gui-menu-new-wizard-translations.diff It was created by some ad-hoc hackery. Strings for languages where I didn't know the encoding were dropped. Is there a standard way to extract the strings in UTF-8 from setup (.lng files) and merge them in .xcu files? The mail says the strings can stay in scp, but that's of course duplication. Would you accept a patch that moves those strings completely to officecfg?
of -> mh: Please have a look.
confirming issue
reassigned.
Both the installation process and the menu configuration will be changed greatly for 2.0. I will keep an eye on these changes, particularly as they relate to configuration, to ensure that they will properly support creation of language packs. The changes that have to be done for this will also include proper migration of the menu texts. I can use this task as a reminder for this. This would mean setting the target to 2.0. For the 1.1 codeline things are more difficult. I don't know the effect of migrating these strings from scp to officecfg on translation databases and gsi files. The change might also create problems for 'patch' upgrades (as opposed to full reinstalls). I am sure that this is not desired for the StarOffice 7 codeline and will create unnecessary extra work for OOo 1.1.x as well. Thus any solution for 1.1 would probably involve: - Keeping the duplicate entries in scp and officecfg - Allow to select either one using a build-time flag The patch that is provided is still far from being feasible: - Of course all existing languages must be migrated properly. For this it will be necessary to write a program/script that extracts the strings from the scp files and converts them to UTF-8. The mapping from language codes to text encodings that is used in setup/scp can be found in some 'lang2textenc' (or similar) function in the tools module (you can find this in LXR, starting from ~ SiConfigurationItem::Execute (setup2/source/agenda/action.cxx)). - To keep the ability to use 'custom setup' (which is why the stuff is in scp in the first place), these settings need special handling in officecfg: The data for the res/<language>/... files needs to be generated, but the corresponding data/... files should not contain the corresponding branch. To achieve this, a special attribute (or 'op' value) needs to be introduced into the officecfg source xcu files and the officecfg/util/alllang.xsl script needs to implement this new mode of splitting up the files. To achieve configurability, a new parameter for the script must be introduced. - Makefiles in scp and officecfg, as well as config_office/configure need to be changed to support that build-time option. Because the given patch is more a sketch of a solution rather than a usable patch, I changed the issue type to ENHANCEMENT for now. I will not be able to do this myself for 1.1 although I am willing to review any patch -> flagging as 'needhelp' Please provide feedback whether you need this for 1.1.x and whether you can help getting this done (properly). If that is not the case I will change the target to 2.0 as explained above.
Will be done for 2.0
Implemented for OOo 2.0 as part of feature #25937# in CWS scp2officecfg.
Source code/build issue: marking as verified
Now available on MWS: closing
.