moz-phab: abort: hidden revision '...' was rewritten as: ...!
Categories
(Conduit :: moz-phab, defect, P2)
Tracking
(Not tracked)
People
(Reporter: nalexander, Unassigned)
References
Details
(Keywords: conduit-triaged)
Attachments
(1 file)
|
122.55 KB,
text/plain
|
Details |
Here's the log:
$ moz-phab submit 6bdb5ac8 4386b834
... (very long backscroll, these commits were very large)
Creating new revision:
573371:4386b834d4c3 Bug 1580356 - TO FOLD - Fix l10n. r?#build
Linting...
No lint engine configured for this project.
Running unit tests...
No unit test engine is configured for this project.
SKIP STAGING Phabricator does not support staging areas for this repository.
Created a new Differential revision:
Revision URI: https://phabricator.services.mozilla.com/D47702
Included changes:
M mobile/android/locales/Makefile.in
rebasing 573352:5848ef0d44b2 "Bug 1580356 - Part 1: Remove Fennec from the build. r?snorp!,#build!"
rebasing 573373:9aa63dd1f897 "Bug 1580356 - TO FOLD - Remove Fennec Java sources. r?snorp"
note: not rebasing 573361:215e33554bc0 "Bug 1580356 - TO FOLD - Remove Fennec Java sources. r?snorp" and its descendants as this would cause divergence
note: not rebasing 573367:4caefae1817d "Bug 1580356 - TO FOLD - Tear out some configure bits. r?#build" and its descendants as this would cause divergence
note: not rebasing 573368:e2240d97a181 "Bug 1580356 - TO FOLD - Tear out docs, some tests, thirdparty references. r?#build" and its descendants as this would cause divergence
note: not rebasing 573369:a797d56f064a "Bug 1580356 - TO FOLD - Tear out API keys. r?#build" and its descendants as this would cause divergence
note: not rebasing 573370:817e5aef98fe "Bug 1580356 - TO FOLD - Fix lints. r?#build" and its descendants as this would cause divergence
abort: hidden revision '9aa63dd1f89702215a1d2e220840bae2d6b11fd3' was rewritten as: e8b9b1740c9f!
(use --hidden to access hidden revisions)
CommandError: command 'hg' failed to complete successfully
Here's what the top of my repository was left as:
Commits:
e8b9b174 o tip draft Bug 1580356 - TO FOLD - Remove Fennec Java sources. r?snorp Nick 2019-09-24
|
81879a64 o draft Bug 1580356 - Part 1: Remove Fennec from the build. r?snorp!,#build! Nick 2019-09-24
|
644f540a | @ draft orphan Bug 1580356 - TO FOLD - Fix l10n. r?#build Nick 2019-09-30
| |
28de3227 | | * draft orphan Bug 1580356 - TO FOLD - Fix lints. r?#build Nick 2019-09-25
| | |
c12280f2 | | | * draft orphan Bug 1580356 - TO FOLD - Tear out API keys. r?#build Nick 2019-09-24
| | | |
52962f17 | | | | * draft orphan Bug 1580356 - TO FOLD - Tear out docs, some tests, thirdparty references. r?#build Nick 2019-09-24
| | | | |
a030e2ed | | | | | * draft orphan Bug 1580356 - TO FOLD - Tear out some configure bits. r?#build Nick 2019-09-24
| | | | | |
8547f260 o | | | | | draft Bug 1580356 - Pre: Don't collect package metrics. r?#build Nick 2019-09-19
| | | | | |
817e5aef | x | | | | draft obsolete Bug 1580356 - TO FOLD - Fix lints. r?#build Nick 2019-09-25
| |/ / / /
a797d56f | x / / / draft obsolete Bug 1580356 - TO FOLD - Tear out API keys. r?#build Nick 2019-09-24
| |/ / /
e2240d97 | x / / draft obsolete Bug 1580356 - TO FOLD - Tear out docs, some tests, thirdparty references. r?#build Nick 2019-09-24
| |/ /
4caefae1 | x / draft obsolete Bug 1580356 - TO FOLD - Tear out some configure bits. r?#build Nick 2019-09-24
| |/
215e3355 | x draft obsolete Bug 1580356 - TO FOLD - Remove Fennec Java sources. r?snorp Nick 2019-09-24
| |
5848ef0d | x draft obsolete Bug 1580356 - Part 1: Remove Fennec from the build. r?snorp!,#build! Nick 2019-09-24
| |
29a4aea8 | x draft obsolete Bug 1580356 - Pre: Don't collect package metrics. r?#build Nick 2019-09-19
|/
6bdb5ac8 o draft Bug 1580356 - Pre: Allow no download information in informulate.py. r?#build Nick 2019-09-11
I see this a great deal with complicated patch stacks, and let me tell you, it's really fun to manually excavate the mountain of rebases left behind.
|
||
Comment 1•6 years ago
|
||
The same issue hit me several times now as well.
|
||
Comment 2•6 years ago
|
||
This seems like Mercurial wanted to rebase a hidden revision.
@sg provided a log. I can't say it's the same issue, as it didn't printed out the error message abort: hidden revision '{}' was rewritten as: {}! nor the note: not rebasing {rev}:{node} "{message}" and its descendants as this would cause divergence
I also can't find the reason for this to happen occasionally.
It was raised in Mercurial.finalize function.
2019-10-16 15:17:24,459 DEBUG $ hg --config extensions.rebase= log -T '{rev} {node}\n' --hidden -r 'successors(38dad706cd4ff74b493bf684383c73fd2e37d0f2) and not obsolete()'
2019-10-16 15:17:24,757 DEBUG 557686 2db76b8b8dfe181d27723eaaab3f72e895f3c407
2019-10-16 15:17:24,758 DEBUG $ hg --config extensions.rebase= log -T '{node}' -r 'parents(e433a6ebf7e49615fd8b50ac52c5431b1e5bfc8b)'
2019-10-16 15:17:24,946 DEBUG
2019-10-16 15:17:24,946 ERROR CommandError: command 'hg' failed to complete successfully
|
|
||
Comment 3•6 years ago
|
||
(In reply to Piotr Zalewa [:zalun] from comment #2)
This seems like Mercurial wanted to rebase a hidden revision.
@sg provided a log. I can't say it's the same issue, as it didn't printed out the error messageabort: hidden revision '{}' was rewritten as: {}!nor thenote: not rebasing {rev}:{node} "{message}" and its descendants as this would cause divergence
The abort message was there, but only on the console, not in the log.
I am not completely sure if I saw the note message.
|
|
||
Comment 4•6 years ago
|
||
The logging is updated.
Please attach the logs if this issue will happen again
Closing due to lack of logs/activity. Don't hesitate to reopen if the issue reoccurs.
|
||
Comment 6•6 years ago
|
||
|
|
||
Comment 7•6 years ago
|
||
Got this today. Log posted in comment 6.
|
|
||
Updated•6 years ago
|
|
||
Updated•6 years ago
|
|
||
Comment 8•6 years ago
|
||
I hit this bug very frequently. Here's the log (no --trace) from just now: https://gist.github.com/mstange/aae5590d6baf80c660c054f82c0d36d7
Roughly, what I did was: Commit A, B, C, D, and then moz-phab submit the range [A, B, C, D]. Then commit E, F, and G, and move E before A using hg histedit, so that the commit order is now E, A, B, C, D, F, G. G is the current working directory. Then moz-phab submit the range [E, A, B, C, D].
|
|
||
Comment 9•6 years ago
•
|
||
Confirmed. MozPhab shouldn't fail like that.
Steps to replicate in Suite (every commit changes different file to avoid conflicts):
echo a > A; hg add A; hg commit -m A
echo b > B; hg add B; hg commit -m B
moz-phab -b 1 -y
echo c > C; hg add C; hg commit -m C
hg histedit
# C < A[D1] < B[D2]
moz-phab -b 1 -y --trace
This results in:
...
rebasing 8:886084132361 "Bug 1 - A"
rebasing 9:b1f6928e442a "Bug 1 - B" (moz-phab_b1f6928e442a)
abort: hidden revision 'b1f6928e442abdddd4f812a874c2d5ec0431abcd' was rewritten as: 4c142195bfcc!
...
The error:
ERROR 2020-07-22 06:16:49,394 Traceback (most recent call last):
File "/home/phab/review/mozphab/subprocess_wrapper.py", line 80, in check_output
output = subprocess.check_output(command, **kwargs)
File "/usr/lib/python3.7/subprocess.py", line 395, in check_output
**kwargs).stdout
File "/usr/lib/python3.7/subprocess.py", line 487, in run
output=stdout, stderr=stderr)
subprocess.CalledProcessError: Command '['hg', '--config', 'extensions.rebase=', '--pager', 'never', '--config', 'rebase.experimental.inmemory=true', 'log', '-T', '{node}', '-r', 'parents(82a3caf96ff3909fb6aefbb3c4634a822c049c0f)']' returned non-zero exit status 255.
|
|
||
Comment 10•6 years ago
|
||
The current way to submit the modified stack would be to submit the parent commit using --single option and then run moz-phab reorg.
In the simplified example (added D to make it necessary to use ranges of commits):
echo a > A; hg add A; hg commit -m A
echo b > B; hg add B; hg commit -m B
moz-phab -b 1 -y
# 4: A[D1] < 5: B[D2]
# A and B are submitted to Phabricator as D1 and D2. D1 is the parent of D2.
echo c > C; hg add C; hg commit -m C
echo d > D; hg add D; hg commit -m D
# 4: A[D1] < 5: B[D2] < 6: C < 7: D
hg histedit
# 8: C < 9: A[D1] < 10: B[D2] < 11: D
moz-phab 8 -b 1 --single -y
# 12: C[D3] < 13: A[D1] < 14: B[D2] < 15: D
# C is now submitted to Phabricator as D3, but not related to D1.
moz-phab reorg 12 14
Which will provide such response:
Reorganisation based on 3 commits:
Stack will be reorganised:
* D1 will depend on D3
Perform reorganisation (YES/No)?
Stack has been reorganised.
|
|
||
Comment 11•6 years ago
•
|
||
The error is in Mercurial::finalize method.
We iterate over each commit in the stack rebasing to get the commits back in order.
Notation: 10:A[D1] == rev:commit_name[Phabricator revision]
Before calling MozPhab the second time (after git histedit) we had the workspce:
7:C[] < 8:A[D1] < 9:B[D1]
During the submission process MozPhab amended the 7:C[] revision to add Phabricator URL to the summary.
In the finalize state (Mercurial::finalize()) the workspace contained hidden commit:
o changeset: 10:C[D3]
|
| @ changeset: 9:B[D2]
| |
| * changeset: 8:A[D1]
| |
| x changeset: 7:C[]
|/
|
o changeset: 0:BASE
MozPhab started to rebase with 8:A[D1]. This moved also the 9:B[D2] and already looked like we want it to be finalized:
@ changeset: 12:B[D2]
|
o changeset: 11:A[D1]
|
o changeset: 10:C[D3]
|
o changeset: 0:BASE
But the iteration https://github.com/mozilla-conduit/review/blob/master/mozphab/mercurial.py#L484-L490 wants to rebase again using the moved 8:B[D2] (hg rebase --source 8:B[D2] --dest 11:A[D1]) which results with an error.
Note: This iteration is required if all of the commits would be amended during the submission process.
Note2: Finalize stage is changing the SHA1 of an already submitted diff.
|
|
||
Comment 12•3 years ago
|
||
I hit this bug again today.
Description
•