Well volume 1 of the 1.0 version has an entire appendix, 11 pages in all. Volume 2 of 1.0 does not seem to have anything on directionality (as 1.0 names it). 2.0 has 10 pages at the end of chapter 3 on Bi-Directional behavior. Typing all this lot up could take a while, and our scanner is broken at the moment (its my wife's, and I must encourage her to go get it fixed as its warrantee will run out soon if she does not get a move on).
David
Shachar Shemesh <wine-devel@sun.consum To: andi@rhlx01.fht-esslingen.de, David.Goodenough@dga.co.uk er.org.il> cc: Wine Devel wine-devel@winehq.com Subject: Re: Old unicode standard 06/15/02 10:35 AM
Andreas Mohr wrote:
Hmm, which approximate time frame is this ? 1998 ? 94 ? 89 ? :-)
I'm talking about the 1991-2 time frame. I believe this makes it ~Unicode 1.0. A later version could come in handy as well. See later on.
David.Goodenough@dga.co.uk wrote:
Shachar,
I have the version 1.0 and 2.0 books, what exactly do you want to know?
In
wha?t form would you like the information
David
I am interested in the BiDi algorythm as defined by these standards. Any format that you can send me is fine. If you would just like to summarize the algorythm, that would be great, but I'm afraid that would amount to huge amounts of work. LIke I said - I would have used the online library, only it doesn't go back that much.
It seems from the information MS released in the MSDN that they are using Version 1.0 of the BiDi algorythm. This goes a great way in understanding why their reordering is so crummy. That leaves us the question of what do we do now.
MS defines only 12 BiDi types. These are not enough to implement the 3.0 BiDi algorythm. I don't know whether the types defined are compatible with the 2.0 algorythm. There is also the question of whether we want to support MS's bugs, in addition to their features.
What version I need depends on what's in the 2.0 standard. If the algorythm there depends on the same 12 types MS defined, I'm leaning torwards implementing that. If it requires further types (as the 3.0 algorythm does), I think I'll go with the 1.0 standard. We may add a special command line feature, available only if a library such as freebidi is available, that will send the strings to an outside library for reordering, which will give us 3.0 support for very little extra work (with performance penalties).
Shachar