You are absolutely right. I can't believe I didn't see that. Tested with no complaints.
I will also update to latest version as noted. Thanks! 2013/3/5 Dan Kennedy <danielk1...@gmail.com> > On 03/05/2013 09:37 PM, Μάριος Φραγκούλης wrote: > >> I have not tapped the VVA_ONLY() definition. >> NDEBUG is included in the compilation options >> I use. Even if it wasn't included, SQLITE_DEBUG is not >> defined so SQLite would define NDEBUG in >> this case (if I understand correctly). >> > > If I understand correctly that's a problem. mem1.zMalloc is > initialized to zero inside a VVA_ONLY() macro. So if NDEBUG > is defined mem1.zMalloc will never be initialized. > > > > >> It may be of interest that I have tapped some >> #if and #ifdef because of compiler complaints. >> For example, a function was called after checking >> that a compile option was defined, but the function >> definition itself would happen either way. Thus the compiler >> complained about unused functions. >> >> Also, I had to wrap a couple of assertions with TESTONLY() or >> VVA_ONLY() (I noticed that this is done elsewhere in the source code) >> because the compiler complained about >> undeclared variables. The assertion that fails is not among those. >> >> I am using SQLite 3.6.22 . >> >> Thanks for taking the time to help me out. >> >> >> 2013/3/5 Dan Kennedy <danielk1...@gmail.com> >> >> On 03/05/2013 07:19 PM, Μάριος Φραγκούλης wrote: >>> >>> Hi All, >>>> >>>> I have compiled SQLite as part of a Linux kernel module >>>> (Linux version 3.2.0.35-generic) >>>> having set SQLITE_OMIT_FLOATING_POINT, >>>> SQLITE_THREADSAFE=0, SQLITE_OMIT_LOAD_EXTENSION, >>>> SQLITE_OMIT_BUILTIN_TEST and NDEBUG. >>>> I have configured SQLite to operate in-memory at all times by setting >>>> :memory: as database name and by turning off the main and temp >>>> journals. >>>> SQLite behaves as expected but /var/log/syslog registers >>>> mem1.zMalloc == 0 assertion failures during queries. >>>> At the assertion point in sqlite3.c it is documented that mem1 is never >>>> malloced, hence the assertion. However, browsing the code >>>> I see various places where zMalloc is indeed set. >>>> >>>> Am I missing sth? >>>> If zMalloc is never malloced, are there any ideas >>>> about the root cause? >>>> >>>> >>> Did the VVA_ONLY() macro get defined correctly? Was NDEBUG >>> defined at build time? >>> >>> ______________________________****_________________ >>> sqlite-users mailing list >>> sqlite-users@sqlite.org >>> http://sqlite.org:8080/cgi-****bin/mailman/listinfo/sqlite-****users<http://sqlite.org:8080/cgi-**bin/mailman/listinfo/sqlite-**users> >>> <http://sqlite.org:8080/**cgi-bin/mailman/listinfo/**sqlite-users<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<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<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
