Subject: Re: i18n of abiword -- OS services
From: Paul Rohr (paul@abisource.com)
Date: Sun Jan 16 2000 - 15:31:51 CST
At 08:51 AM 1/16/00 -0800, Leonard Rosenthol wrote:
> Office97 wasn't fully multi-lingual (ie. it didn't call the
>proper OS routines when it should have). Office2000 no longer has
>that problem - the "stock English" version will work on whatever
>localized OS platform you have (Win98, NT, NT2000).
Is that also true for Win95? I'd gotten the impression from Jeff that it
might be missing some critical APIs (such as ToUnicode, among others).
Given the amount of mail we get from folks who want Windows 3.1 support, I
shudder to think how bad we'd be pilloried for not supporting Win95. :-)
The whole *point* of AbiWord as an alternative is that people shouldn't
*have* to buy a new machine/OS/officesuite package from the Wintel cartel
just to read documents written by someone else who did.
>>If it build in XP so we have one solution for all
>>platform and I think that it should be easier to
>>maintain it.
>>
> However that would only give us support for the specific
>languages that someone is willing to put the effort into (like
>yourself for Thai), while using OS services would give us support for
>ANY localized version of that particular OS.
I think we're all in agreement here. The only question is where and how to
draw the line.
- Above some API boundary, all of the code for i18n and l10n should be
done in XP land, for ease of maintenance.
- Below that API boundary, it'll be great to take advantage of as many OS
services as feasible.
Since the level of services provided by each of our target OSes varies
considerably, we may find ourselves in a situation where the very latest
version of FooOS provides a few services that are "above the line", and thus
duplicating some functionality we already have. So what?
This discussion is most likely to move forward when we have people ready to
code up solutions to some of these specific issues. Their job will be to
write *all* the code needed above the various OS services, and at that point
it shouldn't be hard to evaluate concrete proposals about how high to draw
the XP line.
Paul
This archive was generated by hypermail 2b25 : Sun Jan 16 2000 - 15:26:34 CST