Feature request: flock

[7underlines]

Hi, does v have something like flock from PHP? https://www.php.net/manual/en/function.flock.php
Because I'm looking for a way to prevent double running a long process that is started via cron.
On discord someone mentioned that I could maybe use os.signal, but I'm not so sure how to implement it.

[dumblob]

1. your process shall create a temporary directory (because that is on most OSes atomical unlike creating a file)
2. if it failed (because the directory already existed), quit immediately
3. otherwise (i.e. the directory was successfully created) create a signal handler for SIGINT QUIT HUP ABRT TERM EXIT which deletes the created temporary directory
4. add defer{ remove_my_lock_directory() } in main()
5. nothing else, that's it 😉

[7underlines]

@dumblob Thank you! It's working so far. I've implemented it like this:

fn on_signal() {
	flock_dir := [os.temp_dir(), '.myapp_custom_flock_processing'].join(os.path_separator)
	os.rmdir(flock_dir)
	exit(1)
}

fn main() {
	flock_dir := [os.temp_dir(), '.myapp_custom_flock_processing'].join(os.path_separator)
	os.mkdir(flock_dir) or {exit(0)}

	os.signal(1, on_signal)
	os.signal(2, on_signal)
	os.signal(3, on_signal)
	os.signal(6, on_signal)
	os.signal(15, on_signal)

	// here I do stuff

	os.rmdir(flock_dir)
}

[dumblob]

@logTom glad it works for you. And oh, I forgot to remove the dir at return from main(). I've edited my suggestion above 😉.

[7underlines]

Ah, there is one caveat - this doesn't work if v panics or I stop the program midway with an exit(1) for debugging. I also tried it with defer.

[dumblob]

Ah, there is one caveat - this doesn't work if v panics or I stop the program midway with an exit(1) for debugging. I also tried it with defer.

I'm glad you shout here about this unwise decision. I'm shouting as well. We'll all cry one day about this IMHO biggest issue of V. Read more about panic() being the most harmful feature in V: #2030 (comment) (IMHO this antifeature has the potential to diminish widespread use of V).

On the positive side, defer{} section shall be called irrespectively of panic() if I'm not mistaken. So you're observing a buggy behavior IMHO.

@medvednikov and others, would you "dare" to contemplate about panic() being more and more harmful? In light of channels being currently implemented into the language this'll become a hot topic some day sooner or later anyway.