search

johannes-mueller / dired-gitignore.el

dired minor mode to ignore files that are to be ignored by git
GNU General Public License v3.0
21 stars 3 forks source link

dired-subtree / dired-sidebar integration #11

Closed vasdee closed 1 year ago

vasdee commented 1 year ago

Firstly, thanks for this package, it's a near perfect fit for my setup and as the readme states, alot less heavy than the alternatives.

I've been using dired-gitignore within the excellent dired-sidebar package. One thing I've noticed there is that the use of dired-subtree (I assume) doesn't work well with the filtering of dired-gitignore. My suspicion there is that either dired-get-marked-files doesn't return the files from what dired-subtree contains, or that dired-gitignore was designed for vanilla dired, in that a single folder is displayed at a time, hence the reason for the very flat git check-ignore $(ls -A1) check?

I guess I'm not asking for a fix here, as an emacs hacking noob, just trying to gain an understanding of your approach to dired-gitignore and maybe any advice for me to be able to tackle this myself

johannes-mueller commented 1 year ago

Thanks for your interest.

The problem is that ls -A1 only returns files of the current directory and not its subdirs for git check-ignore to check. In theory you could add the --recursive flag to the ls command. Then git check-ignore would also list all the ignored files of all the subdirs. But that would not really work in practice as then ls would list all the files of the dirtree. And that can be quite a lot these days as people usually have things like python venvs or node_modules or other build dirs with zillion files of all the dependencies of the project in their working directories. And if you feed this whole list to git check-ignore you'd get a argument list too long.

So ls is no longer the right tool to use. I am not aware of a more efficient way to list all the current files that are git ignored in the whole tree. Maybe you could use the fd tool to list the files that are not ignored and then inverse the mark in dired. But we cannot assume that fd is present on the system, so that would be a dependency and would mean hassle for the user.

As always comments and ideas are always welcome.

vasdee commented 1 year ago

Apologies about the late reply, been distracted by a great many things.

I'm actually quite intrigued by using an external tool here. Maybe ls is the fall back but if you have something else, something like silversearcher springs to mind here, then replacing the default with ag -l (or equivalent) is the way to go.

I already use silversearcher extensively within emacs, so at least for me, it wouldn't be a stretch to use it here

vasdee commented 1 year ago

Now I'm just brain dumping here, but maybe the comparison is better against the output of git ls files, rather than checking what is explicitly ignored?

johannes-mueller commented 1 year ago

There is a chance to get this working when we use fd to find all the files that need to be shown (see #13). It seems to perform reasonably. Not working with dired-subtree as of yet, though.

johannes-mueller commented 1 year ago

The problem with dired-subtree is that (dired-goto-file) does find files in subtrees shown by dired-subtree. That would mean that a special mechanism would be needed in case dired-subtree is showing something. While this is probably doable in principle, I am reluctant to implement something to accommodate 3rd party packages.

johannes-mueller commented 1 year ago

dired-subdir aupport is added. No plans to add dired-subtree support as of now. Feel free to reopen if you want to rediscuss or take it on.