- Dec 06, 2019
-
-
Boris Feld authored
-
- Dec 09, 2019
-
-
Pierre-Yves David authored
Septa Unella has been contacted.
-
Pierre-Yves David authored
This is just wrong.
-
- Dec 04, 2019
-
-
Raphaël Gomès authored
-
- Nov 27, 2019
-
-
Raphaël Gomès authored
-
- Nov 29, 2019
-
-
Pierre-Yves David authored
Strictly speaking we could have a variant tracking for "having markers" to clarify some result. In practice: - We don't have a requirement for evolution yet not tracking of stuff like the obstore format - We would need some non-filtered equivalement repository... that does not really exists since we have various draft, orphan and obsolete changesets there - We can still add variants later "without troubles".
-
- Nov 25, 2019
-
-
Raphaël Gomès authored
-
- Nov 05, 2019
-
-
Pierre-Yves David authored
With the new rust code, this is expected to make a difference. TODO: We need a script converting older result to add the (default) CPU count).
-
Pierre-Yves David authored
I suspect we should start gather decorator into a wider "asvbenchmark` one. But this is out of scope of this series.
-
- Nov 08, 2019
-
-
Pierre-Yves David authored
The feature seem common enough that making it accessible globally make sense. We'll introduce a way to easily set it up in the next changesets.
-
Pierre-Yves David authored
Rayon is not respected mercurial configuration in terms of worker. We enforce this through a config variable. (this will become configurable later)
-
- Oct 04, 2019
-
-
Pierre-Yves David authored
Maybe we should split the fast and the slow test in two different jobs ?
-
- Nov 12, 2019
-
-
Raphaël Gomès authored
-
Raphaël Gomès authored
-
- Nov 07, 2019
-
-
Raphaël Gomès authored
parallel cache warm See merge request octobus-public/scmperf!95
-
Raphaël Gomès authored
bench status See merge request octobus-public/scmperf!94
-
- Nov 06, 2019
-
-
Pierre-Yves David authored
If gnu-parallel is available, let us it to speed up cache warming. Currently each benchmark run spend about 5 minutes checking that cache are up to date (they are). Doing so in parallel should at least speed things up a bit.
-
- Nov 07, 2019
-
-
Pierre-Yves David authored
This will be helpful to make sure everything use the same listing and order.
-
- Nov 06, 2019
-
-
Pierre-Yves David authored
-
Pierre-Yves David authored
This will make is simpler to run it in parallel when possible
-
Pierre-Yves David authored
Otherwise the two tests can walk on each other.
-
Pierre-Yves David authored
-
- Nov 03, 2019
-
-
Pierre-Yves David authored
-
Pierre-Yves David authored
In this changesets, we introduced the `hg perfstatus --dirstate` command.
-
Pierre-Yves David authored
We are about to introduce another minimal revision for a new status command. So we need a way to distinguish between the two. The current approach is not great, and we probably need to start thinking about a more generic way to deal with this (probably in a unified `asv_benchmark` decorator). However, in the mean-time lets move forward on status related benchmark.
-
Pierre-Yves David authored
For mercurial, this is close to getting some "hello work" timing, we should have this.
-
- Oct 30, 2019
-
-
Pierre-Yves David authored
We have to update the existing smoke test. We also add a new one dedicated python. It runs a revision that supports python 3.
-
Pierre-Yves David authored
-
Pierre-Yves David authored
It apparently moved in python 3.
-
Pierre-Yves David authored
-
Pierre-Yves David authored
-
Pierre-Yves David authored
Otherwise, access to the relevant data fails.
-
Pierre-Yves David authored
-
Pierre-Yves David authored
-
Pierre-Yves David authored
-
Pierre-Yves David authored
The command output is bytes, we need to feed bytes to the associated operation.
-
Pierre-Yves David authored
The command output is bytes, so the regexp need to be bytes too.
-
Pierre-Yves David authored
It will be used in multiple file system path.
-
Pierre-Yves David authored
This is a file system path, lets make it be bytes.
-
Pierre-Yves David authored
-