code-review
code-review copied to clipboard
Pushes to taskgraph-try have `None` as the commit message.
https://treeherder.mozilla.org/#/jobs?repo=taskgraph-try&author=reviewbot
Indeed, and it's happening in production, not in testing with the latest code.
I got this line in the events logs:
code-review/production/events 2020-02-27 04:06:15.725989 libmozevent.mercurial: Applying patch (phid='PHID-DIFF-bkzfw2z7vbfphj7brpqi' message='None\nDifferential Diff: PHID-DIFF-bkzfw2z7vbfphj7brpqi')
The message is built using the commit message from Phabricator (here is the relevant code)
And here is the result from the Conduit API call on your patch:
{
"data": [
{
"id": 234911,
"type": "DIFF",
"phid": "PHID-DIFF-bkzfw2z7vbfphj7brpqi",
"fields": {
"revisionPHID": "PHID-DREV-mgs6k6xhuweyobdqja6f",
"authorPHID": "PHID-USER-fu2yuh6gonry46x7h3lr",
"repositoryPHID": "PHID-REPO-liwcrizcwtpjejelpbpd",
"refs": [
{
"type": "branch",
"name": "default"
},
{
"type": "base",
"identifier": "6c517d0ed0e152f5ff797cc1824267624a911a42"
}
],
"dateCreated": 1582776366,
"dateModified": 1582776368,
"policy": {
"view": "public"
}
},
"attachments": {
"commits": {
"commits": [
{
"identifier": "d72d54ec3af40ac04e684c5d3f6a63a937272a60",
"tree": null,
"parents": [
"6c517d0ed0e152f5ff797cc1824267624a911a42"
],
"author": {
"name": "Tom Prince",
"email": "[email protected]",
"raw": "\"Tom Prince\" <[email protected]>",
"epoch": 1582776300
},
"message": null
}
]
}
}
}
],
"maps": {},
"query": {
"queryKey": null
},
"cursor": {
"limit": 100,
"after": null,
"before": null,
"order": null
}
}
As you can see, the message
payload is null on your commit d72d54ec3af40ac04e684c5d3f6a63a937272a60
. Which tool did you use to publish on Phabricator ?
I use hg phabsend
from mercurial's phabricator
extension.
I think it would probably make more sense to use the title+summary from the associated revision, as this matches what lando will use when the commit eventually lands (https://github.com/mozilla-conduit/lando-api/blob/master/landoapi/commit_message.py ; although with the replace_reviewer logic from there).