[MDEV-15814] Travis-CI (consistent) failures for 10.3 branch Created: 2018-04-08 Updated: 2018-06-29 Resolved: 2018-06-29 |
|
| Status: | Closed |
| Project: | MariaDB Server |
| Component/s: | Tests |
| Affects Version/s: | 10.3 |
| Fix Version/s: | 10.3.8 |
| Type: | Bug | Priority: | Major |
| Reporter: | Otto Kekäläinen | Assignee: | Teodor Mircea Ionita (Inactive) |
| Resolution: | Fixed | Votes: | 0 |
| Labels: | None | ||
| Environment: |
Travis-CI: Ubuntu 14.04 |
||
| Attachments: |
|
||||||||
| Issue Links: |
|
||||||||
| Description |
|
I noticed that Travis-CI has been failing a lot since January 2018. When I wrote the blog post https://mariadb.org/protected-branches-ensuring-code-quality-git/ there was some sporadic failures, but not permanent failures. Now when I used the commit 1730ac5c4ae1dc62f6fca5e751a8b08e9119be6a as a base and did 4 non-changing commits in a row just to see what results are sporadic or not, I got in each and every build the exactly same failures: Failing test(s): main.type_datetime_hires main.select_pkeycache main.select main.select_jcl6 main.type_float main.func_str main.type_time_hires main.type_timestamp_hires Logs: These happen only in the build with
See screenshot |
| Comments |
| Comment by Otto Kekäläinen [ 2018-04-08 ] |
|
All failures seem to be related to the same floating point being rounded/offset a bit. This is not visible on Buildbot as clang is not tested there. The effect of a small floating point imprecision is not big per se, but the fact that this keeps Travis-CI failing permanently makes this a blocker. This MDEV should be closed either by fixing the issues themselves ( Travis-CI must be become green and kept all green! |
| Comment by Teodor Mircea Ionita (Inactive) [ 2018-06-29 ] |
|
Clang4 has been removed from the Travis build matrix. |