Log handler's unit test were broken by me on June 22, fixing this:
don't auto-log LOGREC_LONG_TRANSACTION_ID in log handler's unit tests, as they read their log and expect to find only the records they asked for, and so not LOGREC_LONG_TRANSACTION_ID. All log handler's unit tests pass. A way to run them faster: when their LONG_BUFFER_SIZE is too big (1GB) I divide it by ten and then I can run them on /dev/shm - they are then "instant". By the way, pushbuild was not displaying anymore "unit: failed" in the list of all pushes, which participates to the reasons why I didn't notice the breakage earlier. The other reason being that I was too lazy to run log handler unit tests on my machine as they took long (hadn't yet thought about the /dev/shm idea) and so I relied on pushbuild; Danny has now quickly fixed pushbuild - thanks :) storage/maria/unittest/ma_test_loghandler-t.c: don't auto-log LOGREC_LONG_TRANSACTION_ID in log handler's unit tests, as they read their log and expect to find only the records they asked for storage/maria/unittest/ma_test_loghandler_multigroup-t.c: don't auto-log LOGREC_LONG_TRANSACTION_ID in log handler's unit tests, as they read their log and expect to find only the records they asked for storage/maria/unittest/ma_test_loghandler_multithread-t.c: don't auto-log LOGREC_LONG_TRANSACTION_ID in log handler's unit tests, as they read their log and expect to find only the records they asked for storage/maria/unittest/ma_test_loghandler_pagecache-t.c: don't auto-log LOGREC_LONG_TRANSACTION_ID in log handler's unit tests, as they read their log and expect to find only the records they asked for
Showing
Please register or sign in to comment