Onliner magic to remove files

glasswalk3r has asked for the wisdom of the Perl Monks concerning the following question:

Greetings monks,

I'm trying to figure out how this onliner to remove files works:

for(<*>){((stat)[9]<(unlink))}

The "*" in the diamond operator is something I can't find any document about it.

Also, calling stat (which will hit $_, filled by the diamond operator of any filename in the current directory) and getting a slice of it (returning mtime of the file as the ninth element of the list) makes no sense to me, specially how the filename information is being given to unlink.

Besides that, calling stat for me makes no sense at all... I could be calling unlink directly, couldn't I? At least it works if I replace everything with an print "$_\n" inside the for loop.

Here is the reference from where I fetched this onliner: http://www.slashroot.in/which-is-the-fastest-method-to-delete-files-in-linux.

Thanks,

Alceu Rodrigues de Freitas Junior
---------------------------------
"You have enemies? Good. That means you've stood up for something, sometime in your life." - Sir Winston Churchill

Comment on Onliner magic to remove files Select or Download Code

Replies are listed 'Best First'.
Re: Onliner magic to remove files (useless) by tye (Sage) on Dec 23, 2014 at 19:57 UTC
It is equivalent to: `for( glob("*") ) { stat $_; # Useless code unlink $_; }` [download] - tye	[reply] [d/l]
Re^2: Onliner magic to remove files (useless) by mr.nick (Chaplain) on Dec 23, 2014 at 20:04 UTC
Thanks, Tye. I've been struggling to figure out what the stat() did myself. I was currently exploring that maybe stat threw some sort of error on an unreadable file, directory or something and that would prevent the evaluation of the right hand expression, but all my tests ALWAYS evaluated both sides regardless. mr.nick ...	[reply]
Re: Onliner magic to remove files by LanX (Saint) on Dec 23, 2014 at 20:07 UTC
See `glob` for the meaning of <*>. Both `unlink` and `stat` operate on $_ which holds the file found by glob. The boolean < is mysterious for me, I can only only guess that a warning is triggered in case of error. Alternatively the evaluation of this boolean to raise the error got lost while copying. Most probably cargo cult. Cheers Rolf _{(addicted to the Perl Programming Language and ☆☆☆☆ :)}	[reply]
Re: Onliner magic to remove files by Anonymous Monk on Dec 24, 2014 at 01:14 UTC
`for(<*>){((stat)[9]<(unlink))}` The apparently useless `stat` and `<` made me a bit curious, so I googled it a bit. There are several places where the above is presented as a one-liner for "fast file deletion", but without explaining what the `stat` might be for. Several people claimed that the Perl solution is faster than find with the `-delete` option, and I found one (possibly dubious) claim that it's because `find` uses `fstatat64` syscalls while Perl uses `lstat64`+`stat64` syscalls (a quick check on my system appears to confirm this). However, since it isn't my area of expertise, I'm not sure if that causes a speed difference. In a post by Kent Fredric - `https://www.quora.com/Linux-why-stat+unlink-can-be-faster-than-a-single-unlink` - he shows that removing the `stat` actually gives a minor performance improvement, so I suspect it really isn't necessary. My best guess for now is that the code snippet is a bastardization of something like this, and it's just been cargo-culted from there.	[reply] [d/l] [select]
Re: Oneliner magic to remove files by Anonymous Monk on Dec 24, 2014 at 19:09 UTC
The `find` based methods in slashroot article are somewhat dubious. Firstly, `-exec rm {} \;` option will invoke rm for each and every file (also noted by the author). Gnu find offers a nifty feature for such occasions. $ time find ./ -type f -exec rm {} \+ real 0m1.764s Which builds up argument list, passing that to rm. It's more or less equivalent to the following (except there's no pipe/xargs involved): $ time find ./ -type f -print0 \| xargs -0 rm real 0m1.321s Using `-delete`: $ time find ./ -type f -delete real 0m0.795s In parallel: $ time find ./ -type f -print0 \| xargs -0 -P4 -n10000 rm real 0m0.571s Perl versions: $ time perl -e 'for(<>){((stat)[9]<(unlink))}' real 0m2.917s $ time perl -e 'unlink for <>' real 0m2.308s Tests were run in a tmpfs backed directory; files created as in the reference article: $ for i in $(seq 1 500000); do echo testing >> $i.txt; done	[reply] [d/l] [select]