Improve virion_infect performance

[Muqsit]

Significantly improves performance of virion_infect by placing changed files in a temporary directory and calling Phar::buildFromDirectory($temp_dir_path). This avoids repackaging phar for every changed file (repackaging happens only during the Phar::buildFromDirectory call).

The complexity is reduced from O(n^2) all the way to O(n) where n is the number of files being packaged into .phar.

[SOF3]

Please make this error-safe. Consider race conditions of path being used by another process between unlink() and mkdir().

[SOF3]

Now you have the chance of two virions building in the same directory and mixing up.

[Muqsit]

I can't think of a way to do this without either throwing an exception (if(!mkdir()) throw $e) or a possibly long-lasting while(!mkdir()) sleep() loop. I thought of going with if(!is_dir()) mkdir() because the only other mkdir call in poggit/poggit goes with that approach, any ideas?

[SOF3]

Choose another tempnam if the directory creation failed. Only run the subsequent code if you are the one who created this directory.

[JaxkDev]

@Muqsit you had a chance to do this ?

[SOF3]

I am not confident that the logic is still correct... If $this->path doesn't already exist, something might have gone wrong, and you should choose another tempnam. If $this->path is already a directory, this also means something is wrong, and you should also choose another tempnam.

[Muqsit]

While I do understand the issue, I unfortunately don't have much knowledge in this field and would prefer someone else take over this task as deleting the wrong directory could lead to unperceivable results, possibly even out of the realms of this program.

[dktapps]

what about sticking getmypid() on the end of the folder name?

[SOF3]

I am not confident that the logic is still correct... If $this->path doesn't already exist, something might have gone wrong, and you should choose another tempnam. If $this->path is already a directory, this also means something is wrong, and you should also choose another tempnam.

[SOF3]

I think the temp dir thing is rare enough to matter. @JaxkDev any other comments?

[JaxkDev]

I think the temp dir thing is rare enough to matter. @JaxkDev any other comments?

I dont touch virions lol

[SOF3]

The complexity is reduced from O(n!) all the way to O(n) where n is the number of files being packaged into .phar.

Are you sure it is O(n!)? It just sounds like O(n^2) to me.

[Muqsit]

The complexity is reduced from O(n!) all the way to O(n) where n is the number of files being packaged into .phar.

Are you sure it is O(n!)? It just sounds like O(n^2) to me.

Yeah. Previously, for each file that was inserted in Phar, n+1 files would be repackaged (n = number of files before the insertion of the new file). I would guess that makes it n!?

[SOF3]

The complexity is reduced from O(n!) all the way to O(n) where n is the number of files being packaged into .phar.

Are you sure it is O(n!)? It just sounds like O(n^2) to me.

Yeah. Previously, for each file that was inserted in Phar, n+1 files would be repackaged (n = number of files before the insertion of the new file). I would guess that makes it n!?

No, that's O(n^2). To be precise, it is. 1 + 2 + ... + n = n(n+1)/2.

O(n!) multiplies them together instead of adding. That would take centuries for something like PocketMine.