Conversation
with those of other modules, like `datalad_next.gitpathspec`.
…ess` with those of other modules, like `datalad_next.gitpathspec`.
All non-deprecated content of this module has been migrated to `datalad_core` (and some to `datasalad`). A deprecation warning it added to the module. All consuming code here has been adjusted to the new imports.
In an effort to trim dependencies on legacy datalad.
| def iter_subproc( | ||
| args: List[str], | ||
| *, | ||
| input: Iterable[bytes] | None = None, | ||
| chunk_size: int = COPY_BUFSIZE, | ||
| cwd: Path | None = None, | ||
| bufsize: int = -1, | ||
| ): |
There was a problem hiding this comment.
There are a few test failures in this PR:
E TypeError: iter_subproc() got an unexpected keyword argument 'input'
I tried to dig into them a bit, and found https://github.com/datalad/datasalad/commit/4b448912ded25e91f9a6c763620d921cb894a8c0, where the "input" parameter of the now used iter_subproc() implementation in datasalad was renamed to "inputs".
I will try to push a change that adjusts to the different parameter name, but I suspect that the problem is that different runners either expect "input" or "inputs"? Let's see...
There was a problem hiding this comment.
sadly, yes, if I adjust the parameter name, while previously failing tests pass, I get tons of
FAILED datalad_next/gitremotes/tests/test_datalad_annex.py::test_submodule_url - datalad.runner.exception.CommandError: CommandError: 'git -c diff.ignoreSubmodules=none -c core.quotepath=false push -u dla dl-test-branch' failed with exitcode 128 [err: 'fatal: TypeError(call_git_oneline() got an unexpected keyword argument 'input')
So this would mean we either homogenize the parameter name across runners (but thus also across several different datalad extensions), revert the renaming in datasalad, or keep the separate runner implementation.
Previously, tests failed with TypeError: iter_subproc() got an unexpected keyword argument 'input' because a previously used implementation of iter_git_subproc had input as a parameter name. The now used implementation in datasalad renamed this parameter to inputs. I'm not sure if this causes other problems down the line because I believe different runners not inconsistently expect either input or inputs...
2 participants
No description provided.