Comment #8 on issue 20799 by [email protected]: Revert bot on Automatic  
tree closure
http://code.google.com/p/chromium/issues/detail?id=20799

Ok, so I implemented a simple revert bot, and I am ready to commit, but I  
have second
thoughts, So please let me know what you think.

My implementation is basically "gcl revert XXX [--commit]", which does the  
revert to
the revertbot server, send the review, and commit if --commit is specified.  
(This is
basically implemented like a try job. "gcl try empty_cl --revision  
REV_TO_REVERT --
bot revertbot". revertbot reverts the change instead of syncing to it like  
the other
try bots do.

What is gives us is:
1. No need to worry about having an up to date tree on your machine
2. really fast
3. You wont commit locally modified files by accident
4. can be automated by the GateKeeper (i.e, GateKeeper sends the command  
without "--
commit", and once a sheriffs notices, he adds the --commit, or he can just  
LGTM the
change... a poller would poll on the review until there is a LGTM, and then  
revert
the change)
5. All commits are from "[email protected]"

What would be hard to add
1. Append a reason why we want to revert to the description. It will just  
say "Revert
XXX, Original description: XYZ", well, I guess someone could edit the  
description on
revert before doing the --commit.


The reason that I have second thoughts is that the main reason we wanted a  
revert bot
was the reverting was a painful task, but with drover, this is not.

drover is already taking care of point 1,2 and 3. It's crazy fast. It also  
lets you
append any notes you want to the description. it's also really easy. You  
don't even
need a checkout at all on your machine.

So, that leaves us with #4 and #5.  #4 is not even implemented right now,  
and might
be more noisy then helpful. Most of the changes breaking the tree don't get  
reverted.
Most of them because they did not need to be reverted... And it's as fast  
to revert
locally with drover than to go to rietveld and do LGTM.

#5 is the key point. It would be great if the changes we revert are not  
under our
name.

Is this enough good reasons to go ahead with the revert bot? (I'm not too  
worried
about the current code, since it took less than 2 hours to write, but  
people will ask
for features, and this would be another piece of infrastructure to  
maintain). Or
should we just teach people how to use drover?  We could even map "gcl  
revert XX" to
"drover --revert XX".

Let me know what you think,

Thanks

--
You received this message because you are listed in the owner
or CC fields of this issue, or because you starred this issue.
You may adjust your issue notification preferences at:
http://code.google.com/hosting/settings

--~--~---------~--~----~------------~-------~--~----~
Automated mail from issue updates at http://crbug.com/
Subscription options: http://groups.google.com/group/chromium-bugs
-~----------~----~----~----~------~----~------~--~---

Reply via email to