>> I don't, however, do any error checking on that eval. > >Jinx!
Heh, heh :)
>> that if a) the eval fails or it gets cached or whatever, then >> b) I'd be returning a blessed object for a module that hasn't >> been loaded. *That* could cause the ->connect to not exist, >> thus causing my error. Right? > >Right.
And this is confirmed further: if I just add "use LibDB::DB::MySQL" to the top of my .cgi scripts (yeah, yeah, the whole "in the same codebase" stuff), then I can't replicate the missing ->connect.
>It's not random; you just don't know what series of steps reproduces it >yet. If you run with -X, you can probably find a sequence that will do
This is what I'm hoping happens (I'm still in the process of building a dev server that exactly mimicks my prod, thus allowing me to run -X). Does the following sound possible?
* the first load of index.cgi uses no DB calls (well, it does "use LibDB::DB", but DB.pm never gets around to eval'ing MySQL.pm), so things are cached without it.
* with all the modules and whatnot cached in the child, a "use the database" call to that process fails because LibDB::DB::MySQL isn't part of that cache. Refreshing like mad eventually brings me to a child that has not previously cached things, and thus, LibDB::DB::MySQL is loaded and cached.
If that's the case, then I should consider eval'ing my database module in the .cgi script, regardless of whether I'm actually using any of it's functions.
-- Morbus Iff ( i put the demon back in codemonkey ) Culture: http://www.disobey.com/ and http://www.gamegrene.com/ Spidering Hacks: http://amazon.com/exec/obidos/ASIN/0596005776/disobeycom icq: 2927491 / aim: akaMorbus / yahoo: morbus_iff / jabber.org: morbus
-- Report problems: http://perl.apache.org/bugs/ Mail list info: http://perl.apache.org/maillist/modperl.html List etiquette: http://perl.apache.org/maillist/email-etiquette.html
