[WIP] : simplify fpath

[agramfort]

attempt to close #712

cc @adam2392 @dmalt

@adam2392 looking more closely at this issue I am not sure what
is the intended difference between .basename and .fpath attribute
in BIDSPath.

CI will be red here but it starts the ball rolling.

@adam2392 you can push to this PR if you want

[hoechenberger]

We could make use of MNE's _validate_type()

[adam2392]

@adam2392 looking more closely at this issue I am not sure what
is the intended difference between .basename and .fpath attribute
in BIDSPath.

IIRC basename is like os.path.basename, whereas fpath is like a pathlib.Path essentially. Furthermore looking at the code...

• basename only gets you the filename (you don't know datatype)
• basename gets you a string object
• fpath gets you a Path object and either the full, or relative path

@adam2392 you can push to this PR if you want

I will try to set aside time soon.

[agramfort]

@adam2392 should basename == BIDSPath.fpath.name ?

[adam2392]

@adam2392 should basename == BIDSPath.fpath.name ?

I believe so

[adam2392]

@agramfort I don't believe I have write access here?

[agramfort]

if you have commit right on mne-bids you do

[hoechenberger]

@agramfort I don't believe I have write access here?

I've just granted you write access to this repo, can you push here now?

[adam2392]

Stupid question: how do I fetch this PR?

(mne) adam2392@Adams-MacBook-Pro mne-bids % git remote  -v
agram   https://github.com/agramfort/mne-bids.git (fetch)
agram   https://github.com/agramfort/mne-bids.git (push)
dev     https://github.com/adam2392/mne-bids.git (fetch)
dev     https://github.com/adam2392/mne-bids.git (push)
upstream        https://github.com/mne-tools/mne-bids.git (fetch)
upstream        https://github.com/mne-tools/mne-bids.git (push)
(mne) adam2392@Adams-MacBook-Pro mne-bids % git fetch upstream pull/721/head:simplify_fpath
fatal: Refusing to fetch into current branch refs/heads/simplify_fpath of non-bare repository
(mne) adam2392@Adams-MacBook-Pro mne-bids % git fetch upstream pull/721/agramfort:simplify_fpath
fatal: couldn't find remote ref pull/721/agramfort

[sappelhoff]

I recommend that you use https://github.com/cli/cli ... it simplifies these sorts of things remarkably well. See the following screenshot (upper right corner in the image):

but for the record, with just git you would do:

git fetch agram simplify_fpath
git checkout simplify_fpath
<do your changes>
git push -u origin my_simplify_fpath
<do a PR to alex's GitHub fork>

If you have maintainer rights (I think you don't right now) you can also push to alex's fork directly ... pretty much what I did for your fork in #717

[hoechenberger]

Stupid question: how do I fetch this PR?

git fetch --all
git switch simplify_fpath

[hoechenberger]

If you have maintainer rights (I think you don't right now) you can also push to alex's fork directly

I've granted @adam2392 write rights, I thought this would suffice … well, let's see!

[adam2392]

My understanding of our simplification is to entirely remove the attempt at "matching basenames" inside fpath, and simply have fpath be an appendage of BIDSPath.directory and BIDSPath.basename. Is this correct?

[agramfort]

My understanding of our simplification is to entirely remove the attempt at "matching basenames" inside fpath, and simply have fpath be an appendage of BIDSPath.directory and BIDSPath.basename. Is this correct?

if this works this would very simple to explain. If you can...

[adam2392]

I am running into an error with the emptyroom stuff, which I am now out of my depth as to how to address the issue. Those functions use fpath to infer some dataset:

FAILED mne_bids/tests/test_path.py::test_find_empty_room - TypeError: expected str, bytes or os.PathLike object, not NoneType
FAILED mne_bids/tests/test_path.py::test_find_emptyroom_ties - TypeError: expected str, bytes or os.PathLike object, not NoneType
FAILED mne_bids/tests/test_path.py::test_find_emptyroom_no_meas_date - TypeError: expected str, bytes or os.PathLike object, not NoneType

@hoechenberger would you be able to assist?

[agramfort]

@adam2392 you need to rebase.

[hoechenberger]

I've resolved the merge conflict. Looking into the emptyroom stuff now.

[hoechenberger]

Moving to 0.10 milestone