Closed Bug 1496166 Opened 7 years ago Closed 7 years ago

ensure arc is fully configured prior to the first submission attempt to avoid leaving repo in an unexpected state

Categories

(Conduit :: moz-phab, defect)

defect
Not set
normal

Tracking

(Not tracked)

RESOLVED FIXED

People

(Reporter: ntim, Unassigned)

References

Details

(Keywords: conduit-triaged)

Attachments

(2 files)

My commits before doing moz-phab submit: - OLDEST central Bug 1491688 - Introduce ::-moz-tree-row(blur) selector. r=emilio Bug 1491688 - Stop applying -moz-appearance: -moz-mac-source-list-selection; when tree is focused. r=jaws - NEWEST After doing moz-phab submit: random commit |_ central - Bug 1491688 - Stop applying -moz-appearance: -moz-mac-source-list-selection; when tree is focused. r=jaws |_ Bug 1491688 - Introduce ::-moz-tree-row(blur) selector. r=emilio Only "Introduce ::-moz-tree-row(blur) selector. r=emilio" was submitted to Phabricator.
There were no errors or anything suspicious in moz-phab's output? Could you paste it, if you still have it?
Flags: needinfo?(ntim.bugs)
Attached file Terminal output
Here's everything I did that day.
Flags: needinfo?(ntim.bugs)
I think ntim's local tree was left in a funny state when arc first failed due to a certificate not being installed. Maybe we should check this first? glob, what do you think? (Also note that ntim terminated the API keys accidentally left in the paste output, so I've left the attachment.)
Flags: needinfo?(glob)
(In reply to Mark Côté [:mcote] from comment #3) > I think ntim's local tree was left in a funny state when arc first failed > due to a certificate not being installed. yeah, it checkout out the first commit in order to submit, which failed. as trying to restore the state of a repo after a failure is complicated and could quite possibly make things worse it's left as-is. > Maybe we should check this first? yes, but it should be once per repository, not on each submission due to the time a conduit call takes: > $ echo '{}' | time arc call-conduit conduit.ping > {"error":null,"errorMessage":null,"response":"ip-172-31-60-226"} > 1.65 real 0.14 user 0.03 sys creating a file in the repo's .hg/.git directory would be the easiest way to make this first-run-only.
Flags: needinfo?(glob)
fwiw: > user@57d3b22625c5:~/firefox $ moz-phab submit > Submitting 1 commit: > 488436:da5e1cd8ad00 Bug 1491688 - Introduce ::-moz-tree-row(blur) selector. r?emilio > Warning: found 16 untracked files: > Submit to Phabricator (YES/No/Always)? No > user@57d3b22625c5:~/firefox $ hg wip > o 488437:db099ce2321b ntim tip central Bug 1491688 - Stop applying -moz-appearance: -moz-mac-source-list-selection; when tree > is focused. r=jaws > @ 488436:da5e1cd8ad00 ntim Bug 1491688 - Introduce ::-moz-tree-row(blur) selector. r?emilio > | o 488435:a6b130810bc4 dao inbound Bug 1491252 - Port URL overflow handling from urlbarBindings.xml to UrlbarInput.jsm. r=standard8 > |/ note 488436 is the active commit; `hg update 488437` was all that was required to put you back to your initial state.
Summary: moz-phab rebases are messing up the commit order and randomly tagging commits as central → ensure arc is fully configured prior to the first submission attempt to avoid leaving repo in an unexpected state
(In reply to Byron Jones ‹:glob› 🎈 from comment #5) > fwiw: > note 488436 is the active commit; `hg update 488437` was all that was > required to put you back to your initial state. 488437 would still have been incorrectly tagged as central though
Blocks: 1484093
We're calling arc after the checkout. If arc fails the repository is in a different state than before calling the `moz-phab`. We should check if `arc` is running correctly before proceeding with the submit. This patch is sending a ping to Phabricator's API. We cache the call by creating a file in the repository dot path.
Status: NEW → RESOLVED
Closed: 7 years ago
Resolution: --- → FIXED
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: