> #test adding config at request time > - my $errmsg = $r->add_config(['require valid-user']); > - die $errmsg if $errmsg; > + $r->add_config(['require valid-user']);
I don't know about this. I was going to work up my thoughts as a result of another thread, but I might as well do so here... I think it's a bad practice for a persistent OO interface to die at runtime. what you've done means that I need to now use eval() logic to handle the exception, rather than simply checking the return value - several extra lines of code (and processing) when 1 will do just fine. and it further obscures our interface: which methods do I need to trap with eval() and which can I trust not to die again? I think it's ok to die at startup or something, but if your request-time handler dies request after request because of, say, a different AllowOverrides config, it's bad. I've followed this rule in my own persistent environments (not only mod_perl :) and found it really to be a best practice. so, I'd suggest re-thinking this a bit. I understand what you were trying to solve, but maybe there is another way? maybe call croak on $s->add_config but not $r->add_config? --Geoff --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
