Hi Otto, On 2020-09-28 9:12 a.m., Otto Kekäläinen wrote: > I am currently working on mariadb-10.5 in Debian and it regressed on > hppa. I would appreciate it if somebody had some fixes to suggest. > > See https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=968915 I don't see that mariadb-10.5 has ever built successfully on hppa. mariadb-10.3 builds okay.
In the last two builds, the mariadb-10.5 build failed due to the following testsuite issue: main.index_intersect w3 [ fail ] Test ended at 2020-09-27 16:41:38 CURRENT_TEST: main.index_intersect --- /<<PKGBUILDDIR>>/mysql-test/main/index_intersect.result 2020-08-07 12:57:35.000000000 +0000 +++ /<<PKGBUILDDIR>>/mysql-test/main/index_intersect.reject 2020-09-27 16:41:38.035115473 +0000 @@ -715,7 +715,7 @@ SELECT * FROM City WHERE Name BETWEEN 'G' AND 'J' AND Population > 500000 AND Country LIKE 'C%'; id select_type table type possible_keys key key_len ref rows Extra -1 SIMPLE City range Population,Country,Name Name 35 NULL # Using index condition; Using where +1 SIMPLE City index_merge Population,Country,Name Name,Country 35,3 NULL # Using sort_intersect(Name,Country); Using where EXPLAIN SELECT * FROM City WHERE ID BETWEEN 1 AND 500 AND Population > 700000 AND Country LIKE 'C%'; mysqltest: Result length mismatch The test passed with 10.3. I can help with tool chain issues but I have not clue on how to debug test issues. The table, ref and Extra fields differ from expected. Is there some reason these might differ for hppa for this test? Was the test changed between 10.3 and 10.5, or something related to test? There is a hppa porter machine available to Debian developers. Helge Deller can also provide access to others on request. Regards, Dave -- John David Anglin [email protected]
