Hi Markus, On Friday, 2011-07-29 00:54:19 +0200, Markus Mohrhard wrote:
can someone from the Thai community check whether these two patches cause any problems. They remove some part of the bug fix for https://bugs.freedesktop.org/show_bug.cgi?id=33089 with patch http://cgit.freedesktop.org/libreoffice/libs-gui/commit/?id=6b5a84d339676d6e34c17cbefbec427e4559f37awhich broke several parts of the number format dialog.
Not from the Thai community ;) but ... Good catch, though it occurs to me that the patch cures only a symptom. First, initialization of bHasValidBracketPrefix should be inside the else if ( lcl_SvNumberformat_IsBracketedPrefix( eSymbolType) ) block, like so else if ( lcl_SvNumberformat_IsBracketedPrefix( eSymbolType ) ) { bool bHasValidBracketPrefix = false; because bracketed prefixes are evaluated for each of the possible 4 sub formats and one sub format can have more than one prefix. With this the actual value of bHasValidBracketPrefix at the end of the switch always equals !bCancel, the prerequisite to execute the final block, so the old code is always executed and the new code else { nPos = nPosOld; // Excel LCID removed } of the malicious change never is. If I didn't overlook some path at some place.. Eike -- PGP/OpenPGP/GnuPG encrypted mail preferred in all private communication. Key ID: 0x293C05FD - 997A 4C60 CE41 0149 0DB3 9E96 2F1A D073 293C 05FD
Attachment:
pgpg8AKdjpwX2.pgp
Description: PGP signature