Closed adswa closed 4 years ago
hey @all-contributors please add @DorienHuijser for bug userTesting
@adswa
I've put up a pull request to add @DorienHuijser! :tada:
I have looked in to this and it doesn't appear to be straightforward :( Neither mimetype
nor mimeencoding
work for for Windows to detect the file type, and textfiles will be placed in the annex (unless consistently saved using save --to-git
). I see the following ways going forward:
text2git
configuration needs to be removed for windows users in any case. We should also add notes to the sections that explain this procedure (at least http://handbook.datalad.org/en/latest/basics/101-124-procedures.html)Apart from that we could either:
save --to-git
. However, one save with --to-git
will place the file into the annex and it will be a pain to get it out of there.Both options are not optimal. I'm leaning towards the first, personally.
Yet another alternative: Go back to recommending the WSL option. But this is, again, suboptimal on so many levels...
After catching up a bit of sleep, I think the easiest solution are a few custom largefile rules inside of DataLad-101
that take the final dataset structure into account:
untested, no internet in train:
*.txt annex.largefiles=nothing
code/** annex.largefiles=nothing
done by #532.
Handbook starts with a text2git configuration, which uses mimeencoding since Dec 2019, but this is not included in git-annex: https://git-annex.branchable.com/todo/provide_windows_build_with_MagicMime/
discovered by @DorienHuijser -- thanks so much!
hey @all-contributors please add @DorienHuijser for bug userTesting