On Thu, Jan 15, 2009 at 10:07:29AM -0800, Brown, Daniel wrote: > The frequent releases are not a problem as far as I am concerned. I'd > rather have bugs fixed quickly when they are discovered, than wait > months for releases containing needed fixes like other libraries. We > use the loose pre-generated C files (not the amalgamation) and even then > it only takes me about 10-20 minutes to integrate a new release and > about another 10-40 minutes to run it through our unit tests.
I agree, but note that I run gmake fulltest. In 3.6.9 gmake fulltest takes many hours (on and amd64 Solaris system it's been running since yesterday, and it's still not finished -- it's currently at savepoint4-2.11.1.7814, and making progress). Is this indicative of a performance regression in 3.6.9? Or just a result of a massively increased number of tests? 3.5.4's fulltest had ~60000 individual tests, 3.6.9's has ~380000, a more than six-fold increment! Incidentally, on Solaris I get the following failures in 3.6.9 gmake fulltest: altermalloc-1.transient.40... Expected: [1 1] Got: [0 {}] attachmalloc-1.transient.40... Expected: [1 1] Got: [0 {}] malloc6-1.transient.40... Expected: [1 1] Got: [0 {}] mallocG-1.transient.40... Expected: [1 1] Got: [0 {}] savepoint3-2.transient.4819... Expected: [1 1] Got: [1 {unable to open database file}] savepoint6-smallcache.294.2... Error: no such table: t1 savepoint6-smallcache.295.1... Error: no such table: t1 ... The savepoint errors went away when I re-ran just those tests, so I think those were a transient issue with my test system. I can't figure out how to run just the malloc tests (putting them in the INCLUDE list in veryquick.test doesn't do it). Should I be concerned? I'm also seeing ticket #3205, but that's a problem with Tcl and not a show stopper for me. Nico -- _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users