I am so confused

[dpc]

I run:

> lychee .

it reports checking 239 links. OK, let's try just md files.

> ls **/*.md | wc -l
44
# OK, as expected
> ./result/bin/lychee '**/*.md'
# starts spinning, and is already at 1363

Whaaa? OK, let's try verbose mode to figure why this is happening.

> ls **/*.md | wc -l
44
> ./result/bin/lychee -vvvvvv '**/*.md'
# output slightly different, but not much difference or information anyway

OK... I --help, maybe --dump-inputs will help. Yeap. This shows that a lot of links comes from ./direnv, but why? It's in .gitignore already. I add it to .lycheeignore, no change.

Edit:

OK, so .lycheeignore is for urls that I want to ignore. Great, very useful.

But I still don't understand the behavior of selecting inputs.

[mre]

In the current implementation, globs don't respect .gitignore.
We use the glob crate, which doesn't provide this functionality.

The code is here:
https://github.com/thomas-zahner/lychee/blob/76dbd843f4b3360821153df1685914ba08682e07/lychee-lib/src/types/input.rs#L298-L303

We considered switching to globset, but it also wouldn't respect .gitinore files. (It would allow multiple glob patterns, though!)
I am not aware of any other crate which respects the .gitinore file, but we could implement it ourselves.

[dpc]

@mre Oh. Thanks. That explains it.

I just quickly checked what typos is doing and it seems to be using ignore which says:

The ignore crate provides a fast recursive directory iterator that respects various filters such as globs, file types and .gitignore files.

Sounds exactly like what you're looking for?

[mre]

Interesting. I wasn't aware that ignore supports globs. We use it already, so we could remove our custom glob handling and replace it with ignore. If you have the time, I'd appreciate a pull request to fix the issue. All good if not.

[dpc]

Sorry, I am a type of user that complains and does nothing. ;D . Most of the time at least.

[mre]

Heh, all good. Now is the time to change that. 😂