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

I need to generate a unique identifier to a large number of users accessing my code. The unique identifier has to be the number of seconds passed since 00:00::00 2000 GMT.

I am able to get the date from the time::local module but how would i mess around with this date without having to go through loads and loads of convert to seconds modules. I am using my $time = gmtime(time)

THis also must be dynamic so that i dont have to upgrade the code every time the calendar year changes.

thanks, cos

Replies are listed 'Best First'.
Re: time in seconds
by Hofmator (Curate) on Aug 08, 2001 at 16:59 UTC

    not sure if I'm getting what you mean ... but what about simply my $seconds = time - 946684800; time gives the seconds since 00:00:00 UTC, January 1, 1970 and subtract from this the seconds till 00:00::00 2000 UTC.

    -- Hofmator

[reply]
[d/l]
Re: time in seconds
by mattr (Curate) on Aug 08, 2001 at 17:16 UTC
    The sequence of "number of seconds passed since x" will not provide enough unique ids for a "large number of users" i.e. >1 user per second.

    How about $t = time-$deltasecs . "|" . int(rand(10e5));

    overkill: Apache::Session::Generate::MD5

[reply]
[d/l]
Re: time in seconds
by scain (Curate) on Aug 08, 2001 at 17:18 UTC
    While Hofmator is right, you may want to look at Date::Calc which allows many time and date transformations done easily.

    Scott

[reply]
Re: time in seconds
by thpfft (Chaplain) on Aug 08, 2001 at 18:04 UTC

    Small confusion to clear up: time returns seconds-since-epoch (ie 1970). gmtime() converts seconds-since-epoch into either a readable string or an array of hours and minutes and so on depending on whether it's called in scalar or list context. Performing backflips to parse a gmtime(time) string into seconds would not be effort well spent.

    So the seconds counter is easy, as Hofmator shows. But it probably isn't good enough: two people could very easily use your script during the same second if, as you say, there are a large number of them.

    Here are a few other options to consider:

    seconds . ip address;   # if on web. has anonymity implications
seconds . username;     # if command-line. also not anonymous
$ENV{UNIQUE_ID};      # if in apache & compiled with mod_unique_id
use Time::HiRes;     # microseconds.

    [download]

    or then there's Apache::Session, which would be overkill to start with but might save you some days later.

[reply]
[d/l]
Re: time in seconds
by costas (Scribe) on Aug 08, 2001 at 18:24 UTC
    I have decided to use the epoch time in second and then use it as the seed value in the rand function.

    i am therefore doing

    my $id = int(rand($epochdate));

    Does anybody know what kind of calculation is being done using this rand function and if there is a possibility of two end results being the same (having been generated from the same epoch time or from different time)

    costas
[reply]
[d/l]
      This approach does not work because it is unpredictable when a previously assigned number shows up.
      What about using a simple (persistent) counter without using time or rand at all?

      MP

[reply]

Back to Seekers of Perl Wisdom