Thank you, I have backported it to chromium as http://src.chromium.org/viewvc/chrome?view=rev&revision=55504
Now, how about fts1 and fts2? The original chromium patch is at http://codereview.chromium.org/174387 . Could you take a look and suggest a way to upstream those fixes to SQLite? On Fri, Aug 6, 2010 at 12:10, Richard Hipp <d...@sqlite.org> wrote: > FTS3 updated here: http://www.sqlite.org/src/ci/b8b465ed2c > > On Fri, Aug 6, 2010 at 2:24 PM, Scott Hess <sh...@google.com> wrote: > > > This bug comment describes the problem: > > http://code.google.com/p/chromium/issues/detail?id=15261#c20 > > > > excerpt: > > > Apparently the problem is caused by tolower(), whose behavior is > affected > > by current > > > locale. Under locale tr_TR.UTF-8, tolower('I') returns 'I' rather than > > 'i', because > > > lower case of 'I' defined in tr_TR is 'ı' (U+0131). > > > > I think at the time the bug was being diagnosed, sqlite3_strnicmp() > > wasn't being exposed. I think that does the right thing because it > > uses the internal UpperToLower table. > > > > -scott > > > > > > On Fri, Aug 6, 2010 at 11:11 AM, Richard Hipp <d...@sqlite.org> wrote: > > > If "ch" is an unsigned char then how is the following unsafe: > > > > > > ch = (ch<0x80) ? tolower(ch) : ch > > > > > > And why does it need to be changed to > > > > > > ch = (ch>='A' && ch<='Z') ? ch - 'A' + 'a' : ch; > > > > > > There is only one such instance of code remaining in FTS3 (at > > > fts3_tokenizer1.c:196) but I want to understand what the issue is > before > > I > > > change it. > > > > > > On Fri, Aug 6, 2010 at 1:30 PM, Paweł Hajdan, Jr. > > > <phajdan...@chromium.org>wrote: > > > > > >> On Wed, Aug 4, 2010 at 15:23, Paweł Hajdan, Jr. < > > phajdan...@chromium.org > > >> >wrote: > > >> > > >> > I'm attaching a suggested patch to fix locale-unsafe usage of > tolower > > in > > >> > FTS code. The goal is to make Chromium closer to the upstream, so if > > you > > >> > have a better solution, that's great. > > >> > > >> > > >> Oh, I have just noticed that the mailing list removes all attachments. > > What > > >> is the best way to send patches then? > > >> > > >> By the way, any suggestions about the Chromium patch I linked to ( > > >> > > >> > > > http://src.chromium.org/viewvc/chrome/trunk/src/third_party/sqlite/safe-tolower.patch?view=markup > > >> )? > > >> It seems that it has somehow been fixed in fts3 code. I'm not yet very > > >> familiar with the SQLite codebase though, so could you point me to the > > >> fixes? > > >> _______________________________________________ > > >> sqlite-users mailing list > > >> sqlite-users@sqlite.org > > >> http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users > > >> > > > > > > > > > > > > -- > > > --------------------- > > > D. Richard Hipp > > > d...@sqlite.org > > > _______________________________________________ > > > sqlite-users mailing list > > > sqlite-users@sqlite.org > > > http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users > > > > > _______________________________________________ > > sqlite-users mailing list > > sqlite-users@sqlite.org > > http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users > > > > > > -- > --------------------- > D. Richard Hipp > d...@sqlite.org > _______________________________________________ > sqlite-users mailing list > sqlite-users@sqlite.org > http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users > _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users