- Oct 22, 2019
-
-
Raphaël Gomès authored
`taskset` is linux-only, while this supports unix, mac and windows. As a side-effect, the build phases will *not* be set to only CPUs 2 and 3, but will take advantage of the full power of the machine.
-
- Sep 25, 2019
-
-
Pierre-Yves David authored
New files, new polutions !
-
- Oct 18, 2019
-
-
Pierre-Yves David authored
A new format variant has arrived, we need to take it in account.
-
- Sep 26, 2019
-
-
Pierre-Yves David authored
Bash -x mode issue a lot of confusing output, we remove it and issue a bit more output ourself. note: the test output work with mercurial up to mercurial-core-ba5b062a1388 (included).
-
- Oct 08, 2019
-
-
Pierre-Yves David authored
This cover various status mode. XXX There is a `perfstatus` command that we might be interesting in.
-
Pierre-Yves David authored
-
Pierre-Yves David authored
-
- Oct 04, 2019
-
-
Raphaël Gomès authored
main-run: add the LOG file to hgignore See merge request octobus-public/scmperf!79
-
Pierre-Yves David authored
A new requirement appears, we have to take this in account.
-
- Oct 01, 2019
-
-
Pierre-Yves David authored
We no longer run publish, so these output are not expected anymore.
-
Pierre-Yves David authored
launch: remove the final asv publish See merge request octobus-public/scmperf!78
-
- Sep 26, 2019
-
-
Pierre-Yves David authored
Not sure why this started to happened, but seems relatively harmless.
-
- Sep 25, 2019
-
-
Pierre-Yves David authored
We get pull this one out of the graveyard and put in back into play.
-
Pierre-Yves David authored
having an unknown HGMODULEPOLICY will crash mercurial, effectively preventing it to work before the introduction of the rust module policy.
-
- Sep 24, 2019
-
-
Pierre-Yves David authored
The asv service write information about its work within the LOG file. So we need it properly ignored so that is does not appears in `hg status` as unknown.
-
Pierre-Yves David authored
The publish is dealt with in the main-run.sh now.
-
Pierre-Yves David authored
We don't need stderr output about the unknown command.
-
scmperf authored
It makes the repository having changes which makes the following deploys to fails.
-
Pierre-Yves David authored
-
- Sep 16, 2019
-
-
Raphaël Gomès authored
-
- Sep 24, 2019
-
-
Pierre-Yves David authored
The cache warming stage for larger repository have make this overflow
-
Pierre-Yves David authored
Let's try to move toward a new a better naming.
-
Pierre-Yves David authored
This will make it simple to pick a benchmark name.
-
Pierre-Yves David authored
If things are slow, we'll do with less loop. That is better than very long bench run and no numbers from time out.
-
Pierre-Yves David authored
We should probably do that upstream
-
Pierre-Yves David authored
And try our best otherwise.
-
Pierre-Yves David authored
-
Pierre-Yves David authored
-
Pierre-Yves David authored
-
- Sep 23, 2019
-
-
Pierre-Yves David authored
They have a dedicated "identical" benchmark now.
-
Pierre-Yves David authored
-
Pierre-Yves David authored
We expect these to make the rust code shine.
-
Pierre-Yves David authored
All other function name have "discovery" in them. Let us be consistent here.
-
Pierre-Yves David authored
-
Pierre-Yves David authored
-
Pierre-Yves David authored
We don't want error to silently pass, revision should either be silently skipped or work.
-
Pierre-Yves David authored
-
Pierre-Yves David authored
This is very useful when a benchmark are recenty introduced but is compatible with a larger range of revision.
-
Pierre-Yves David authored
This decorator that getting into play "after" the setup. That was too late.
-
Pierre-Yves David authored
We use a clear, simple but direct approach to skip test involving http for problematic revision.
-