Le 04/12/12 21:24, Tyler Romeo a écrit :
> Don't we have some sort of policy about an individual merging commits that
> he/she uploaded? Because these three changes:
> https://gerrit.wikimedia.org/r/36801
> 
> 
> https://gerrit.wikimedia.org/r/36812
> 
> https://gerrit.wikimedia.org/r/36813
> 
> Were all uploaded and submitted in a matter of minutes by the same person,
> and each is a fix for errors in the commit before it. It kind of defeats
> the point of having code review in the first place.

Overall there is a lot of issues in this class such as :

 - shelling out to possible non existent command (slooow)
 - using file locks (everytime you want a new UID it writes a file, lock
it, generate the uid, unlock the file and delete it.  That seems slow to
me and probably not going to scale.
 - There is ton of code duplication when the number of bits we want
should be a parameter to a generic function.

uniqid( ) will probably give you what you want without having to shell
out.  It is based on microtime() but you could add more entropy by
passing a string prefix as first arg and true as second arg to add in
pseudo random value.


Anyway that looks like a work in progress, I have submitted a change to
revert the commits from master:

 https://gerrit.wikimedia.org/r/36961


-- 
Antoine "hashar" Musso


_______________________________________________
Wikitech-l mailing list
[email protected]
https://lists.wikimedia.org/mailman/listinfo/wikitech-l

Reply via email to