is this okay?

[siliconmike]

pseudocode:

function generate_unique _id()
{
mkdir("lock");

if successful
{
$id = 10 digit random number;
make sure that $id does not exist in our table
if it does, generate a new one

insert $id into our table;

rmdir("lock");
}
else try mkdir again
}

do you see any race conditions?

this idea could be dangerous if the script died before rmdir("lock")..
any solution?

[I can't use auto-increment .. long story]

Mike

[Iván Sánchez Ortega]

Re: is this okay?

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

siliconmike wrote:
[color=blue]
> mkdir("lock");[/color]

Why don't you use the flock() mechanism instead??


- --
- ----------------------------------
Iván Sánchez Ortega -i-punto-sanchez--arroba-mirame-punto-net

Un hombre con pereza es un reloj sin cuerda.- Balmes.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.2 (GNU/Linux)

iD8DBQFDWhrc3jc Q2mg3Pc8RAubuAJ 9UpWw+7wqBftaSQ +Q+QXZ07tC7ZgCf RXI0
hQFGVcJJc7zopnu pH4160dc=
=gGrZ
-----END PGP SIGNATURE-----

[Andy Hassall]

Re: is this okay?

On 21 Oct 2005 17:58:41 -0700, "siliconmik e" <siliconmike@ya hoo.com> wrote:
[color=blue]
>pseudocode:
>
>function generate_unique _id()
>{
> $id = 10 digit random number;
> make sure that $id does not exist in our table
> if it does, generate a new one
>
> insert $id into our table;[/color]

You mention a "table", so I assume this is going into a database.

If so, simply add a unique constraint to the column, and leave the locking to
the database.

Don't check if it exists with a SELECT since that introduces a race condition;
attempt the INSERT, and catch the unique key violation if there is one.
--
Andy Hassall :: andy@andyh.co.u k :: http://www.andyh.co.uk
http://www.andyhsoftware.co.uk/space :: disk and FTP usage analysis tool