On 8/12/2015 10:19 PM, Rik Kabel wrote:
On 2014-03-04 11:07, Khaled Hosny wrote:
Hi Hans,
sorry for telling again, but I’m afraid that the following sample shows a serious buggy interaction between \setupdirections[bidi=on] and \about references.
\setupdirections[bidi=on] No problem with \setupdirections[bidi=on,method=two] (which I still
On Mon, Mar 03, 2014 at 07:49:07PM +0100, Pablo Rodriguez wrote: think should be the default, but Hans is not convinced).
Regards, Khaled
Khaled,
I too had to rely on method=two as well to get around the problem with references appearing at the bottom left of some pages, but this seems to be resolved with the recent betas and the changes Hans has made at Talal's urging.
Perhaps you could test to see if resolves your cases.
there are three methods: - default : based on how idris uses structured tex + mixed latin/arabic - one : from khaled, optimized and extended a bit by me - two : a follow up on 'one' by me (updated unicode bidi specs + more clever fences) 'default' is more efficient and ok for most mixed cases like the ones idris needs; i will probably leave 'one' as-is but will at some point optimize method 'two' and probably add some options there; also, in the case of for instance writing manuals 'default' might be a better choice anyway, the problem with all methods (and more if it's a strict one) is that (as with more solutions) it's hard to get around when you want something different; also unicode bidi evolves (and actually the last version handles some cases that were the reason for 'default' in the first place as idris noted conceptual issues with unicode bidi at that time) so one can wonder about compatibility issues with any such mechanism that applies some heuristics to input instead of tagging now, messing with the text stream is one aspect of bidi but that will not solve layout issues and special cases like hyperlinks, margin notes, images and such because simply reversing the par dir (or text dir) can interfere a lot (the dir mechanism in luatex is taken from omega and was never made for real complex mixed cases) .. many mechanisms need to be made r2l aware so ... in order to get everything all right we need to test and improve stepwise ... no other way (i'm playing with a better handling of line number for instance, where we have quite some interaction between paragraph properties and putting numbers at the right spot, and that's only one thing of many ... and i prefer to leave the luatex engine untouched ... i'm not really a user of r2l myself so ... a slow free-time process) Hans ----------------------------------------------------------------- Hans Hagen | PRAGMA ADE Ridderstraat 27 | 8061 GH Hasselt | The Netherlands tel: 038 477 53 69 | voip: 087 875 68 74 | www.pragma-ade.com | www.pragma-pod.nl -----------------------------------------------------------------