- From: Aryeh Gregor <Simetrical+w3c@gmail.com>
- Date: Fri, 30 Jan 2009 14:00:45 -0500
- To: fantasai <fantasai.lists@inkedblade.net>
- Cc: www-style@w3.org
On Fri, Jan 30, 2009 at 12:41 AM, fantasai <fantasai.lists@inkedblade.net> wrote: > Does the algorithm have errors, or just the examples? The description given in the spec definitely is correct for 1-999. For 0 it's as correct as anything (I don't think there's any other way to write zero if it's to be supported at all). For numbers 1000 and up, there are multiple correct ways to do it; I don't know if all the details of the way given there are correct, or what all of the other ways to do it are. Traditionally not many numbers this large came up in practice (and when they did you could just write out the words), while in modern times the decimal system is widely used in Israel in addition to the Hebrew system. > Also, can you produce a diff against Overview.src.html? > CVS instructions are available if you know CVS: > http://dev.w3.org/cvsweb/ > > Otherwise, if you can give me exact code to paste into > http://dev.w3.org/csswg/css3-lists/Overview.src.html > or just an updated copy of that file, then I can fix it > that way. Actually, now that I'm reviewing the examples again, I think I see that it's saying something different from what I thought it was. The examples accord with the explanation that was intended to be given, and my algorithm does not. I was interpreting the phrase "one for each group up to the zero" as meaning "one for each consecutive group of zeros", when it actually meant "one for each group after and including the zeros, regardless of contents". I actually know Hebrew and misunderstood it, so the wording could probably be improved somehow. I'm thinking of how this could best be achieved. An alternative would be to arbitrarily specify a maximum representable number of, say, 999,999. This would drastically simplify the description for basically no practical loss in utility. The standard could note that larger numbers could potentially be represented, but that UAs may use decimal in those cases (as they're forced to with many other algorithmic systems). Do you/anyone else think this would be a good idea?
Received on Friday, 30 January 2009 19:01:21 UTC