Re: No syntax text properties!

Andy> You can ask Stephen, Because the patch is broken. I've already identified a bug and a suspicious regression in 21.4.9 Windows [3]. make check is broken. That's fixed in my workspace; all the failed tests were due to not conditioning on the syntax-table feature, as you would expect. But it needed to be checked. Based on inspecting the patch, I thought maybe "\\<" in regexps broke too, but it tests OK. Writing up the test case for this also turned up a bug in the regexp "\\b" (which should be treated the same as "\\<\\|\\>" but isn't). That bug is present in vanilla 21.4, too. I've got a couple more unexplained regressions to check, and a bunch of code cleanups I want to do since the new bug makes it certain that I'll be looking at that code in the near future. Then I'll commit to the 21.4 mainline. Andy> but the default branch and windows branch are virtually Andy> identical now. You shouldn't have any problems using the Andy> windows branch under UNIX. Agreed. Expect 6 benign syntax test failures in make check, everything else should be the same. In stock 21.4.9, (setq lookup-syntax-properties nil) may be enough. In TraverseSchema.java from Xerces (thank you very much for that, Stef!) that setq exhausts the speedups for me. In fact, a stock 21.4.9 build with all error-checking turned on indents reportSchemaError as fast as a 21.4.9 Windows [3] with all error-checking off! But the stock build is 6 times as slow with (setq lookup-syntax-properties t). YMMV. Andy sees more speedups with Windows [3] compared to the setq. -- Institute of Policy and Planning Sciences http://turnbull.sk.tsukuba.ac.jp University of Tsukuba Tennodai 1-1-1 Tsukuba 305-8573 JAPAN My nostalgia for Icon makes me forget about any of the bad things. I don't have much nostalgia for Perl, so its faults I remember. Scott Gilbert c.l.py