use Perl

• How's it look without disk caching? (Score:2)

  by Elian (119)

  I'd bet a good part of your performance is coming from the OS caching the file in memory. How's the performance look if there's not nearly so much cache handy? (Still good, I expect, but...)
  • Re:How's it look without disk caching? (Score:2)

    by Matts (1087)

    I'm no good at that sort of low level thing... How would I find out?
    • Re:How's it look without disk caching? (Score:2)

      by Elian (119)

      The most sure way is to bounce the machine before doing the read test. That's probably a bit much, so the next best way would be to try the read test again after you're pretty sure that the disk caches have been flushed. Running Mozilla for a while's a good way to do that. :)
• Fork & Threads (Score:0)

  by djberg96 (2603)

  How does SQLite handle fork? Can a child inherit a database handle or will it choke (like most RDBM's)?

  Also, have you tried using Perl's threads to run concurrent queries? If you don't have a threaded Perl, there's always the "Inline" modules. :)

  Just curious.
  • Re:Fork & Threads (Score:2)

    by Matts (1087)

    Well it locks the entire DB on updates anyway, so I wouldn't recommend it much for multi-user systems.
    
    I'm curious about all that stuff too, but it's hard to setup those sorts of tests. Well actually not *that* hard - I could just set it up under mod_perl, but I don't have that many hours in the day ;-)
• Third party support (Score:1)

  by lachoy (1663)

  The next version of SPOPS, which will come out whenever I wake up -- I have a strict no-bleary-eyed CPAN release policy -- includes support for SQLite. (Oracle too, but let's keep on message.)

  One thing I noticed is that SQLite doesn't seem to like table definitions with an explicit 'NULL' declaration. For instance, the following will fail:

  CREATE TABLE testme (
  id int not null primary key,
  name varchar(20) null
  )

  with a DBD::SQLite::db do failed: near "null": syntax error at ...

  Other tha
  • Re:Third party support (Score:2)

    by Matts (1087)

    Watch out for the memory leaks though :-D
    
    I've got a fix for some of them, but I'm still seeing slight leaks. I'm wondering if that's not in the underlying library though, since leaking 1 scalar per request would see more leakage than I've seen so far.
    • Re:Third party support (Score:1)

      by lachoy (1663)

      I'm glad I checked the latest CPAN releases before releasing this version of SPOPS -- I see you put the $dbh->last_insert_rowid method in, which means that I don't have to use the random string key generation method :-)