[MDEV-5272] [PATCH] MTR/mysqltest overlays for included files do not work on Windows Created: 2013-11-10 Updated: 2013-11-11 Resolved: 2013-11-11 |
|
| Status: | Closed |
| Project: | MariaDB Server |
| Component/s: | None |
| Affects Version/s: | 10.0.5, 5.5.33a |
| Fix Version/s: | 5.5.34, 10.0.6 |
| Type: | Bug | Priority: | Minor |
| Reporter: | Elena Stepanova | Assignee: | Elena Stepanova |
| Resolution: | Fixed | Votes: | 0 |
| Labels: | tests | ||
| Attachments: |
|
| Description |
|
Lets say we have a suite foo, which has a test bar.test, which calls source qux.inc. On Linux, it works as expected, when foo-X.bar is executed, the file basedir/X/mysql-test/foo/qux.inc is used in the source command. But on Windows it is not so – basedir/mysql-test/suite/foo/qux.inc is still called. I got it work by this change: http://bazaar.launchpad.net/~elenst/maria/5.5-elenst-overlays/revision/3945 , but I'm not 100% sure it's the right and complete fix. In the revision above, only a change in mysqltest.cc is the fix, the rest is a test case for the problem. It modifies mtr2.single test so that it calls overlay.inc, and introduces overlay.inc files in the base mtr2 suite and in the MyISAM overlay for it. It also adds single.rdiff file for MyISAM. So, the expected result is when the base suite is executed, it calls the base overlay.inc which executes select 2, which matches the result file. When the MyISAM overlay is executed, it should call its own overlay.inc which executes select 3, which matches the rdiff file. On Windows, without the fix the test fails:
Which shows that the included file is wrong (but rdiff is picked up all right). |
| Comments |
| Comment by Sergei Golubchik [ 2013-11-10 ] |
|
Ok to push, but please move your added lines below, line 3036, directly above "bool in_overlay" declaration. Currently you're doing it in the wrong place, because test_if_hard_path() won't work on converted paths, as far as I can see. |
| Comment by Elena Stepanova [ 2013-11-11 ] |
|
Pushed into maria/5.5: http://bazaar.launchpad.net/~maria-captains/maria/5.5/revision/3945 storage_engine tests in buildbot indicate that the problem has been fixed: (there are failures but they are irrelevant to this report). |