Paul McCullagh <paul.mccullagh@primebase.org> writes:
mysqltest: At line 3: query 'alter table t1 rename mysqltest.t1' failed: 1025: Error on rename of './test/t1' to './mysqltest/ t1' (errno: 16)
Occurs because the PBMS code is compiled in.
This can be disabled by commenting out:
#define PBMS_ENABLED
in xt_defs.h
[xt_config.h actually]
This is, in fact, a bug in PBMS. I have already reported the bug here: https://bugs.launchpad.net/pbms/+bug/416969
So, if the PBMS_ENABLED code is disable (which should probably be the default at the moment), then that test should run through.
Ok, thanks, that solved 3 of the 4 remaining failures. And the last one is just a missing result file update for test pbxt.ps_1general. You have this diff in the last commit for ha_pbxt.cc: - stats.data_file_length = ot->ot_table->tab_rec_eof_id; + stats.data_file_length = xt_rec_id_to_rec_offset(ot->ot_table, ot->ot_table->tab_rec_eof_id); So the data file length output of show table status was fixed from units of records to units of bytes. I just updated the result file accordingly, from 1 to 1024.
The other problems seem to have to do with case-sensitivity. I will see if I can repeat those errors on one of my machines.
Yes, they are case sensitivity issues (except udf maybe). As I said, they all pass with the patch I gave, which just adds missing *-master.opt files that are in the main test suite, but where not copied into the pbxt suite along with the *.test files. So I will merge this into MariaDB with attached patch to disable PBMS and fix test failures. I really want to also enable the PBXT test suite in our Buildbot. Unfortunately I am really pressed for time do do this now. For one, I see a number of Valgrind errors when running the suite that I need to investigate. But one of these days, Thanks, - Kristian. -- https://code.launchpad.net/~paul-mccullagh/maria/maria-pbxt-rc2/+merge/10303 Your team Maria developers is subscribed to branch lp:maria.