This is an archive of the discontinued Mercurial Phabricator instance.

test: backout update evolution config
AbandonedPublic

Authored by lothiraldan on Sep 21 2017, 6:52 AM.

Details

Reviewers
pulkit
quark
Group Reviewers
hg-reviewers
Summary

The concept is still call "Changeset Evolution". The "troubles" got renamed to
"instability", so the action can be renamed from "evolve troubles" to
"stabilize instability". However, the concept itself is still "Changeset
Evolution" so associated configuration should still live under "evolution".
Revert 6c1a9fd8361b first then the config renaming.

Diff Detail

Repository
rHG Mercurial
Lint
Lint Skipped
Unit
Unit Tests Skipped

Event Timeline

lothiraldan created this revision.Sep 21 2017, 6:52 AM
pulkit accepted this revision.Sep 21 2017, 7:48 AM
pulkit added a subscriber: pulkit.

The evolution config name makes sense. stabilization was ambiguous.

quark requested changes to this revision.EditedSep 21 2017, 3:18 PM
quark added a subscriber: quark.

Although "stabilization" may be controversial, but I don't think "evolution" is a better choice.

Since we are here, I think we should come up with formal names. It makes more sense to me to split the giant config into different smaller ones, like:

[experimental]
obsstore = bool
allowunstable = bool
obsexchange = bool
...

That's at least more consistent with expermental.allowdivergence, and is more flexible.

The reason experimental.stabilization is bad can not result in the conclusion that experimental.evolution is good. So @pulkit's +1 is questionable. Therefore requested changes to remove this from 2nd Pass queue. Feel free to request review again to move it back to 1st Pass queue if you believe this patch is still a good approach.

This revision now requires changes to proceed.Sep 21 2017, 3:18 PM

I concur with @quark: if we're going to bother renaming this again, we should just use the [experimental] namespace.

Sorry for the lag, I was at a conference.

I didn't planned for such changes, but I think it's a very good idea! I'm gonna send patches soon. Do you think I should start a new series of patch or add new patch on top of the backout?

I would bias towards just having one change that advances to using [experimental] if we want to rename this again. I don't think we need to take the rollback-then-forward approach, seems like unnecessary ceremony. Let's just talk this over at the sprint?

@durin42 Good for me, I don't have strong opinion on the matter. I will have a patch ready for the sprint.

@durin42 Good for me, I don't have strong opinion on the matter. I will have a patch ready for the sprint.

What's the story with this revision? Should we close it and D762? Or do we have some revised versions I should look at?

lothiraldan abandoned this revision.Oct 16 2017, 12:48 PM

I've started splitting the configuration, need to finish it before the freeze. As the stack will be unrelated to D761 and D762, I think we can close this stack.