The following thread is being moved from xemacs-patches. Please keep Torsten in the CCs, until Michael checks in, Torsten is the expert. A few comments from the immoderator: 1. Please don't continue threads on xemacs-patches, even if others do. I'd like to put a 'bot in place, but until then we depend on your cooperation. 2. Please do use report-emacs-bug. It puts a lot of junk in, most of it is not relevant, but it's better to have junk most of the time than not have the information on the once a week or more occasion when it would be helpful. It's especially nice when professionals do it; then the average user will look and say, "oh, that's how the pros do it." 3. 21.4.6 is in prerelease. See the XEmacs-beta archives for the announcement. 4. I don't know about recent versions, but as of Win NT 3.51 it was possible to ask the OS to allocate a segment of 0xFFFFFFFF bytes, and it would give it to you, wrapping right around 0. People on the DJGPP list used to use this to implement fast graphics. I saw that and ran for Linux as fast as my legs would carry me. No, "NT" was not a typo. Never used that exploit myself, but it was confirmed by trustworthy observers. As for the bug itself, I've queried the person responsible for the patch implementing the submatch argument (which was introduced by the GNU Emacs project several years after we put something else there :( ). It's quite possible I screwed up the application somehow in an early release. Thanks to Torsten's analysis and patch it seems highly likely we should have a fix in time for 21.4.6. We may be able to catch some other bugs due to bogus dereferences, too. It seems pretty evil to move a dereference that might segfault before the test to determine whether it will segfault :-( Bad GCC, bad bad bad GCC! Subject: Topics Topics: Fixing evil replace-match Re: Fixing evil replace-match Re: Fixing evil replace-match Re: Fixing evil replace-match -- Institute of Policy and Planning Sciences http://turnbull.sk.tsukuba.ac.jp University of Tsukuba Tennodai 1-1-1 Tsukuba 305-8573 JAPAN Don't ask how you can "do" free software business; ask what your business can "do for" free software.