Am 26.06.2007 um 21:25 schrieb Dov Feldstern:
Stefan Schimanski wrote:+ // If the last logical character is a separator, skip it, unlessI thought it's about the visually last (or first for rtl paragraphs) character. Take+ // it's in the last row of a paragraph + if (end > 0 && end < par.size() && par.isSeparator(end - 1)) + skipped_sep_vpos = bidi.log2vis(end - 1);HEBREW[ english] rendered as [ english]WERBEH No?I also thought so, but then I realized that it's more subtle. The real problem is with the last *logical* position, which is usually, but not always, at one of the ends. See the attached document for an example. (This is the other manifestation of the bug that I mentioned last night). Try it with and without the patch...
Ok, after rereading your previous mail I got it. Would be good to put a better documentation there like:
// Spaces at line breaks in bidi text can appear visually in the middle // of a row and must be skipped during painting: // * logically "abc_[HEBREW_\nHEBREW]" // * visually "abc_[_WERBEH\nWERBEH]" Haven't tested the patch, but it looks good. Stefan
PGP.sig
Description: Signierter Teil der Nachricht