porcelain.status() fails when submodule directories are present

[teward]

Hello.

There is a small issue with Dulwich and it's Porcelain.status() functions. I'm working on a git project that has submodules and some other folders, and I get some failure cases with porcelain.status() (this is from a Python 3 shell):

Python 3.5.2 (default, Nov 17 2016, 17:05:23) 
[GCC 5.4.0 20160609] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> from dulwich.repo import Repo; from dulwich import porcelain
>>> repo = Repo('.')
>>> porcelain.status()
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/usr/local/lib/python3.5/dist-packages/dulwich/porcelain.py", line 743, in status
    unstaged_changes = list(get_unstaged_changes(r.open_index(), r.path))
  File "/usr/local/lib/python3.5/dist-packages/dulwich/index.py", line 567, in get_unstaged_changes
    blob = blob_from_path_and_stat(full_path, os.lstat(full_path))
  File "/usr/local/lib/python3.5/dist-packages/dulwich/index.py", line 545, in blob_from_path_and_stat
    with open(fs_path, 'rb') as f:
IsADirectoryError: [Errno 21] Is a directory: b'./ChatExchange'
>>> porcelain.status(repo)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/usr/local/lib/python3.5/dist-packages/dulwich/porcelain.py", line 743, in status
    unstaged_changes = list(get_unstaged_changes(r.open_index(), r.path))
  File "/usr/local/lib/python3.5/dist-packages/dulwich/index.py", line 567, in get_unstaged_changes
    blob = blob_from_path_and_stat(full_path, os.lstat(full_path))
  File "/usr/local/lib/python3.5/dist-packages/dulwich/index.py", line 545, in blob_from_path_and_stat
    with open(fs_path, 'rb') as f:
IsADirectoryError: [Errno 21] Is a directory: b'./ChatExchange'
>>> import os
>>> os.getcwd()
'/home/teward/VCS Repos/GitHub/teward/SmokeDetector'

Why it's checking a directory that is a submodule, I'm not sure. Is this a limitation of the 'porcelain' system, and ultimately not the proper way to get current git status?

[jelmer]

On Mon, Apr 17, 2017 at 01:52:33PM -0700, Thomas Ward wrote: Traceback (most recent call last): File "<stdin>", line 1, in <module> File "/usr/local/lib/python3.5/dist-packages/dulwich/porcelain.py", line 743, in status unstaged_changes = list(get_unstaged_changes(r.open_index(), r.path)) File "/usr/local/lib/python3.5/dist-packages/dulwich/index.py", line 567, in get_unstaged_changes blob = blob_from_path_and_stat(full_path, os.lstat(full_path)) File "/usr/local/lib/python3.5/dist-packages/dulwich/index.py", line 545, in blob_from_path_and_stat with open(fs_path, 'rb') as f: IsADirectoryError: [Errno 21] Is a directory: b'./ChatExchange' >>> import os >>> os.getcwd() '/home/teward/VCS Repos/GitHub/teward/SmokeDetector' ``` Why it's checking a directory that is a submodule, I'm not sure. Is this a limitation of the 'porcelain' system, and ultimately not the proper way to get current `git` status?

At the moment the porcelain status code indeed does not support submodules, which is why this is happening. If you look in dulwich/index.py, a couple of lines up from where this error is triggered (line 565) you'll see a TODO comment about this. The issue tracking support for submodules in porcelain is #506 Jelmer

[teward]

@jelmer Your message never got back to me, sorry. Is it not possible to check if there's a submodule and if so "skip" that directory? While that's not the easiest thing to do, surely somewhere in git repo data and configuration that information is stored, and we can exclude submodule items from checks.

[jelmer]

On Thu, May 04, 2017 at 06:54:20AM -0700, Thomas Ward wrote: @jelmer Your message never got back to me, sorry. Is it not possible to *check* if there's a submodule and if so "skip" that directory? While that's not the easiest thing to do, surely somewhere in git configuration that information is stored, and we can exclude submodule items from checks.

That is possible, but it should not be that much harder to support submodules properly in status.