.travis.yml: add pair of linux-ppc64le targets. #6192
Conversation
|
It's about 2 minutes provisioning time. There is no support ccache-ing. There is ccache, but cache is not offloaded, so there is no gain. As result non-extended test build is couple of minutes slower than it could have been. One can also observe several minutes variation, as if jobs are scheduled to less and more busy systems, unbalanced that is. Extended test can take >30 minutes (if you happen to run on apparently more busy system). This is >2x slower |
|
Extended test adds between 30 to 50[!] minutes of additional CPU time. I'd say this is a bit excessive. I therefore would rather suggest to keep one non-extended --strict-warning build and keep sanitizer build commented for occasions when there is reason to believe that there is PPC-specific problem that can be troubleshooted with sanitizer... Removing WIP... |
dot-asm
changed the title
.travis.yml
Outdated
| @@ -109,6 +113,12 @@ matrix: | |||
| - gcc-mingw-w64 | |||
| compiler: x86_64-w64-mingw32-gcc | |||
| env: EXTENDED_TEST="yes" CONFIG_OPTS="no-pic" | |||
| # Uncomment if there is reason to believe that PPC-specific problem | |||
| # can be diagnosed with with possibly >30 mins sanitizer build... | |||
There was a problem hiding this comment.
" can be diagnosed with with possibly " ?
One is clang --strict-warnings and one gcc sanitizer extended test. --strict-warnings build masks couple of warnings and intention is to address them separately. Sanitizer build is quite expensive, can take >30 mins and is commented for occasions when there is reason to believe that PPC-specific problem can be diagnosed with sanitizer.
Idea is to have them started earlier, so that non-extended test doesn't have to be waited out each time.
|
Re-based and ready to roll. |
richsalz
added
the
approval: done
One is clang --strict-warnings and one gcc sanitizer extended test. Sanitizer build is quite expensive, can take >30 mins and is commented for occasions when there is reason to believe that PPC-specific problem can be diagnosed with sanitizer. Reviewed-by: Rich Salz <rsalz@openssl.org> (Merged from #6192)
WIP, no approval requested. Second commit that removes most targets will be removed prior approval is requested. The request will be closed and reopened couple of times in order to get some timing statistics.