sl ghstack crashes if remote bookmark ancestor isn't pulled locally

[discentem]

I cannot tell from this crash message why there was a crash. I wonder if it's because I don't have write access to the repo but not 100% sure.

% sl ghstack
** Sapling SCM (version 0.1.20221118-210929-cfbb68aa) has crashed:
Traceback (most recent call last):
  File "/opt/homebrew/Cellar/[email protected]/3.8.15/Frameworks/Python.framework/Versions/3.8/lib/python3.8/site-packages/edenscm/__init__.py", line 106, in run
    dispatch.run(args, fin, fout, ferr, config)
  File "/opt/homebrew/Cellar/[email protected]/3.8.15/Frameworks/Python.framework/Versions/3.8/lib/python3.8/site-packages/edenscm/dispatch.py", line 161, in run
    status = (dispatch(req) or 0) & 255
  File "/opt/homebrew/Cellar/[email protected]/3.8.15/Frameworks/Python.framework/Versions/3.8/lib/python3.8/site-packages/edenscm/dispatch.py", line 503, in dispatch
    ret = _runcatch(req)
  File "/opt/homebrew/Cellar/[email protected]/3.8.15/Frameworks/Python.framework/Versions/3.8/lib/python3.8/site-packages/edenscm/dispatch.py", line 712, in _runcatch
    return _callcatch(ui, req, _runcatchfunc)
  File "/opt/homebrew/Cellar/[email protected]/3.8.15/Frameworks/Python.framework/Versions/3.8/lib/python3.8/site-packages/edenscm/dispatch.py", line 721, in _callcatch
    return scmutil.callcatch(ui, req, func)
  File "/opt/homebrew/Cellar/[email protected]/3.8.15/Frameworks/Python.framework/Versions/3.8/lib/python3.8/site-packages/edenscm/scmutil.py", line 147, in callcatch
    return func()
  File "/opt/homebrew/Cellar/[email protected]/3.8.15/Frameworks/Python.framework/Versions/3.8/lib/python3.8/site-packages/edenscm/dispatch.py", line 683, in _runcatchfunc
    return _dispatch(req)
  File "/opt/homebrew/Cellar/[email protected]/3.8.15/Frameworks/Python.framework/Versions/3.8/lib/python3.8/site-packages/edenscm/dispatch.py", line 1218, in _dispatch
    ret = runcommand(
  File "/opt/homebrew/Cellar/[email protected]/3.8.15/Frameworks/Python.framework/Versions/3.8/lib/python3.8/site-packages/edenscm/ext/undo.py", line 146, in _runcommandwrapper
    result = orig(lui, repo, cmd, fullargs, *args)
  File "/opt/homebrew/Cellar/[email protected]/3.8.15/Frameworks/Python.framework/Versions/3.8/lib/python3.8/site-packages/edenscm/ext/journal.py", line 80, in runcommand
    return orig(lui, repo, cmd, fullargs, *args)
  File "/opt/homebrew/Cellar/[email protected]/3.8.15/Frameworks/Python.framework/Versions/3.8/lib/python3.8/site-packages/edenscm/ext/sparse.py", line 504, in _tracktelemetry
    res = runcommand(lui, repo, *args)
  File "/opt/homebrew/Cellar/[email protected]/3.8.15/Frameworks/Python.framework/Versions/3.8/lib/python3.8/site-packages/edenscm/ext/copytrace.py", line 173, in _runcommand
    return orig(lui, repo, cmd, fullargs, ui, *args, **kwargs)
  File "/opt/homebrew/Cellar/[email protected]/3.8.15/Frameworks/Python.framework/Versions/3.8/lib/python3.8/site-packages/edenscm/dispatch.py", line 913, in runcommand
    ret = _runcommand(ui, options, cmd, d)
  File "/opt/homebrew/Cellar/[email protected]/3.8.15/Frameworks/Python.framework/Versions/3.8/lib/python3.8/site-packages/edenscm/dispatch.py", line 1259, in _runcommand
    return cmdfunc()
  File "/opt/homebrew/Cellar/[email protected]/3.8.15/Frameworks/Python.framework/Versions/3.8/lib/python3.8/site-packages/edenscm/dispatch.py", line 1217, in <lambda>
    d = lambda: util.checksignature(func)(ui, *args, **strcmdopt)
  File "/opt/homebrew/Cellar/[email protected]/3.8.15/Frameworks/Python.framework/Versions/3.8/lib/python3.8/site-packages/edenscm/util.py", line 1285, in check
    return func(*args, **kwargs)
  File "/opt/homebrew/Cellar/[email protected]/3.8.15/Frameworks/Python.framework/Versions/3.8/lib/python3.8/site-packages/edenscm/ext/ghstack/__init__.py", line 51, in ghstack_command
    return submit_cmd(ui, repo, *args, **opts)
  File "/opt/homebrew/Cellar/[email protected]/3.8.15/Frameworks/Python.framework/Versions/3.8/lib/python3.8/site-packages/edenscm/ext/ghstack/__init__.py", line 104, in submit_cmd
    ghstack.submit.main(
  File "/opt/homebrew/Cellar/[email protected]/3.8.15/Frameworks/Python.framework/Versions/3.8/lib/python3.8/site-packages/ghstack/submit.py", line 162, in main
    base = GitCommitHash(sh.git("merge-base", f"{remote_name}/{default_branch}", "HEAD"))
  File "/opt/homebrew/Cellar/[email protected]/3.8.15/Frameworks/Python.framework/Versions/3.8/lib/python3.8/site-packages/ghstack/sapling_shell.py", line 50, in git
    return self._run_sapling_command(["log", "-T", "{node}", "-r", f"ancestor(., {remote})"])
  File "/opt/homebrew/Cellar/[email protected]/3.8.15/Frameworks/Python.framework/Versions/3.8/lib/python3.8/site-packages/ghstack/sapling_shell.py", line 87, in _run_sapling_command
    stdout = self.sh(*full_args, env=env)
  File "/opt/homebrew/Cellar/[email protected]/3.8.15/Frameworks/Python.framework/Versions/3.8/lib/python3.8/site-packages/ghstack/shell.py", line 218, in sh
    raise RuntimeError(
RuntimeError: /opt/homebrew/bin/sl log -T {node} -r ancestor(., dev) failed with exit code 255

[shish]

RuntimeError: /opt/homebrew/bin/sl log -T {node} -r ancestor(., dev) failed with exit code 255

What happens if you try running /opt/homebrew/bin/sl log -T '{node}' -r 'ancestor(., dev)' by itself, are there any more helpful messages there?

[bolinfest]

If you don't have write access to the repo, then sl ghstack will definitely not work, though if that is what is causing this output, we should certainly communicate that better.

[discentem]

Hey folks sorry for the late reply!

@shish running it manually sort of gets a better error message but still unclear what it's trying to do? I guess it's trying to pull request with a branch that doesn't exist?

% /opt/homebrew/bin/sl log -T '{node}' -r 'ancestor(., dev)'
abort: unknown revision 'dev'!

[discentem]

@bolinfest it turns out write access is not the issue. as a test I got myself write access and still ran into the same issue

[discentem]

Okay I made some progress, but still not solved. The suggestion run the command manually really helped!

So it's looking for dev which is not a revision/bookmark I have locally. Okay fine...

sl pull -B dev

but then I run into another fun error:

The commit 5050b9a6e0d129dc12767b7a51e0e4f96e85681d has 2 parents, which makes my head
explode.  `git rebase -i` your diffs into a stack, then try again.

[discentem]

There are a couple of issues with this error message though.

• I can't use git in a sapling checkout. I guess this message is an artifact of using ghstack.
• sl rebase -i asked me to specify a destination but it's not very clear what destination I should choose.

[discentem]

Just in case it's helpful for someone to understand/explain the issue to me, here's my sl tree state: https://gist.github.com/discentem/482939c9250296b2ad8be17f1e46ea93. Not sure how to get out of this.

[discentem]

I was able to rebase and get further with sl ghstack. However, because I had initially messed up my local state, I decide to reclone and try to pull in remote state of the stack by cloning the head branch and ran into this issue:

RuntimeError: This commit is poisoned: it is from a head or base branch--ghstack
cannot validly submit it.  The most common situation for this to
happen is if you checked out the head branch of a pull request that was
previously submitted with ghstack (e.g., by using hub checkout).
Making modifications on the head branch is not supported; instead,
you should fetch the original commits in question by running:

What's the correct way to import (e.g. pull down) the remote state of the stack?

[discentem]

I ended up solving this by uncommitting the commit that was based on head and re-committing: this seems to satisfy ghstack.

[discentem]

Submitted https://github.com/facebook/sapling/pull/340 to mitigate https://github.com/facebook/sapling/issues/290#issue-1471716614