Hey @jez, I love this program, can you review this PR pls?

I changed print! statements for write!(BufWriter, ...) for a small performance improvement.

By doing some benchmarks using hyperfine, I saw improvements from 3% up to 8%, they were smaller than I expected, but I kept the changes anyway.

So yeah, I tried to improve at other parts but couldn't get better results.

Hi I just recently discovered this tool (via fd) and I think it works great! I just needed an option to display the full path like the -f option in tree does.

Hi

This is not an issue but just a query. I just recently came across this repository and was excited to test it. However, there doesn't appear to be any Windows binaries release I could take out for a spin. I briefly tried installing via rust/cargo but quickly ran into some errors. Is there support (or at least a plan) for Windows ?

https://github.com/jez/as-tree/blob/01253c2e5c3c624802d2246967fce7d73ed1e044/.travis.yml#L24

Is it a good idea to store this token in .travis.yml? Wouldn't it be better to use an environment variable which you can store on travis?

Hey,

I created this project a landing page using Gitpages, which you can watch here: https://twitter.com/gitpages/status/1415339126887522310

Here is a demo of the page in action: https://www.gitpages.app/examples/as-tree.html

Just wondered if you would like the page on the project or not?

Up to you.

You can find out how to use GitHub Pages to host the file if you wanted to: https://docs.gitpages.app/docs/guides/hosting#step-2---setup-github-pages

Would love to know your thoughts or feedback!

Cheers, Dave

I saw you attempted this a few weeks ago?

If you want you can take this and adapt it however you want.

(ps: this is not meant to be merged, just to signal)

Hi,

I have an old linux distribution that does not come with GLIBC_2.18. So the linux binaries archive does not work for me.

Would it be possible to add a release target that either compiles with GLIBC statically? Or alternatively, a MUSL target, which seems to be a static build anyway, similar to ripgrep or fd ?

I realy like the feature but it's strength would potentiate if you could use any character as an separator. If you would e.g. print out JSON as dot-separated string (https://github.com/tomnomnom/gron) you could than view the whole thing as an tree

When finding paths with find and fd, I often pipe the results into a script I wrote called pathsort that groups files before directories.

Piping the output of pathsort into as-tree has no effect, as as-tree sorts its input.

It would be nice if as-tree had an option to group files before directories.

Gnu ls has the --group-directories-first option. tree has the --dirsfirst option. A long flag like those would be fine, but a one letter short flag would be ideal. Neither has an option to group files first.

For list output, I can see both groupings to be useful, but the existing tools only offer directories first.

For find and tree output, for some reason grouping files before directories makes more sense to me. I'm not sure why. Maybe it's because once you've seen the files in the directory, you can be done thinking about that directory for the remainder of the output.

This is my pathsort script:

#!/bin/sh
set -euo pipefail

# 1.  Replace the last `/` with `/ /`
# 2.  Sort
# 3.  Replace the `/ /` with `/`
pathsort()
{
        sed -E 's/(.*)\//\1\/ \//' | sort | sed -E 's/(.*)\/ \//\1\//'
}

case "${1-}" in
        -fd )
                # It would be nice if `fd` had an option to prepend `./` to all relative paths,
                # so that all paths have at least one `/` in them.
                sed 's/^/.\//' | pathsort
                ;;
        * )
                pathsort
                ;;
esac

Off topic

I have this in my shell's rc file:

fdtree() { fd . -H -E .git "$@" | as-tree ; }

I've been using fdtree more than tree lately because I can write fdtree -E aaa -E bbb to exclude using multiple patterns, while tree can only have one -I ignore pattern.

Thanks for as-tree. I love it!

Fixes a panic when writing to a closed pipe by explicitly ignoring this error type.

This commit also changes the behavior of as-tree on empty inputs which would previously result in a (superfluous) newline.

closes #15 closes #19

If fdfind folder_name returns a empty output | as-tree prints a new-line. Is it possible to not make it print new-line?

Actual behavior:

$ fdfind folder_name  # returns no solution
$ fdfind folder_name | as-tree                                       

$

Wanted behavior:

$ fdfind folder_name  # returns no solution
$ fdfind folder_name | as-tree                                       
$