Moved issue gerrit:5309 to now be  issue chromium:683252 .

This exists as a plugin. Which project are you referring to specifically? Or do you mean all projects with a CQ?

"chromium/tools/build" specifically. But I suspect that all projects with a CQ will have a similar issue.

chromium/tools/build has the following chumpdetector config:
{
  "treeName": "Infra",
  "viewURL": "https://infra-status.appspot.com/",
  "statusURL": "https://infra-status.appspot.com/current?format\u003djson",
  "withCredentials": false,
  "enforceCommitQueue": true
}

the final entry ("enforceCommitQueue": true) means that it should throw up a warning if you click the Submit button.

Have you actually observed it not doing this? Especially within the last couple days since the new chumpdetector config was put in place?

Not in the last few days that I can recall.

Here's a trivial CL for you to approve and then one of us can hit the submit button and see if we get the popup: https://chromium-review.googlesource.com/430906

Yeah looks like it pops up. Maybe still address the "Submit" terminology? "Send to CQ"?

Long term, we plan to hide the actual Submit button from everyone except folks who are either sheriffs or grant themselves break-glass permissions through another system. In the mean time, I'd prefer to keep the "Submit" terminology because that's what the CQ uses in the background, it's what Gerrit calls it when calculating whether or not to display the button, etc.

That's fine. I am just pointing out that I've observed several accidental submits over the last few weeks. Those can have real impact (unfortunately).

Yep. If you can remember other repos that have had accidental submits, we can audit their chumpdetector configs as well to make sure they enforce the commit queue.

This just happened again to "tools/build": https://chromium-review.googlesource.com/c/430155/

Changing the title since there is a pop-up warning, and it's still happening in spite of that.

I think this bug should just track the problem of people accidentally hitting "Submit". The consequences of this are high enough that we should spend time making sure it never happens.

Brainstorming: maybe the pop-up can have a "I understand this will bypass CQ" checkbox?

Sana: did you bypass the pop-up? If so, did you do so on purpose, or just click through without reading it? Or did you not see the pop-up at all? And if so, are you using the old non-polymer UI?

If Sana didn't *accidentally* bypass the CQ, there's no new information here.

FWIW, when I was testing it, there was a popup that showed up. I tried clicking it on a CL, and I think that it appeared, and then the CL submitted anyways, before I had a chance to click anything. Which sounds broken. I tried to reproduce, but couldn't.

Also, the author of that CL said that she didn't see any sort of popup when she clicked the button (that she remembers).

(meant to post this last night, but somehow it didn't work)

RE #13, sorry, not trying to be a badger here, just trying to represent the problem.

This number needs to be zero for infra repositories, since one bad submit can cause havoc (e.g., an errant character in "common/chromium_utils.py" will take all of our baremetals offline). That's obviously a separate problem in and of itself, but currently the CQ is our best defense :(

Re #13: I'm not the author of https://chromium-review.googlesource.com/c/430155.

However I did intentionally bypass CQ for https://chromium-review.googlesource.com/c/430155/ and I'm assuming martiniss intentionally bypassed the CQ for https://chromium-review.googlesource.com/c/433087/ as well.

dnj@: Right, if there's a problem, we definitely need to fix it.

My point is that both https://chromium-review.googlesource.com/c/430155 (submitted by smut@) and https://chromium-review.googlesource.com/c/433087/ (submitted by martiniss@) were manually submitted, bypassing the CQ, *on purpose*. They are not good demonstration CLs for "the UI is confusing". Both smut@ and martiniss@ decided that they wanted to bypass the CQ and then did so. The UI worked appropriately in both cases.

Totally agreed that the incidence of people accidentally submitting and bypassing the CQ should be zero. Totally willing to listen to feedback on how to make that less likely. Haven't seen any evidence that it is actually happening, since we put the popup in place.

The CL that prompted #12 is: https://chromium-review.googlesource.com/#/c/430155/

I was told that katthomas@ accidentally hit "Submit to CQ", causing a problem, so I updated this bug with that info. I don't know what the other CLs in #17.

cl/155153151 moves the Submit button into the overflow, reducing confusion

The above change has been landed but not yet deployed. Marking fixed, can verify after deployment.