[xxbondsxx]

Yep, this is where our regexes start to fall short. Mind opening a github issue for this? We can likely add another option in our "fallback" regex, but it's a fine line to draw between overmatching and undermatching.

For instance, we currently match on "add/remove" output from git status which is less than fortunate. For performance reasons we don't actually go check if each of those files exists (since sometimes you want to pipe hundreds of results into fpp [EDIT: and are using an expandrive or network mount])

[ploxiln]

    [pierce@plo-pro team15]$ time (find . -print0 | xargs -0 stat -f '%N' | wc -l)
        4901
    
    real	0m0.079s
    user	0m0.043s
    sys	0m0.060s

[xxbondsxx]

@ploxiln -- edited my comment explaining a lot of devs here use network mounts (in which case checking for existence takes a lot longer). I'd love to explore this though, either behind a config setting or eventually as the default behavior! What do you think?

[sandyarmstrong]

I think to have usable matching you're going to have to go that way, unless you want to write custom parsing code for each command's output.

[ploxiln]

yeah, if the files are not either local or on an NFS mount with sub-millisecond latency (my test took ~ 1 second over NFS to a VM), it's going to take too long. I hadn't thought of that situation

[jankassens]

You could test for a single file (possible with a timeout) and see if it's "fast enough" to check all files.

You could also check in the background and update the display if the heuristic turned out wrong.