Re: Potential problem in 4.3.3RC3

On Sat, 16 Aug 2003, Phil Driscoll wrote: > I mentioned the following issue when 4.3.3RC1 was released. Perhaps the > general feeling is that it is a non-issue, but I think it might at least > cause a lot of support traffic on the lists. > > When I tested SquirrelMail on my test server (reporting set to E_ALL) I get a > load of 'headers can't be sent' messages. The reason is that on 4.3.2 and > it's predecessors, it was ok to make multiple calls to session_start(). > However, from 4.3.3RC1 onwards, every call after the first one generates a > warning message. > > So on production machines with appropriate error_reporting levels, nobody is > going to notice, but on E_ALL machines all hell breaks loose because headers > then can't be sent. I suspect this will affect many more applications than > just SquirrelMail. > > I apreciate that issuing the warning is proably the technically correct thing > to do, but given that there don't seem to be many practical advantages in > issuing the warning, should we revert to the old behaviour? It's a bug in SquirrelMail, so I don't see why we should remove the warning. The warning is to show that you might be doing things wrong, thus it should stay in. And because it's only an e_notice, I don't think it's a problem anyway. Derick -- "Interpreting what the GPL actually means is a job best left to those that read the future by examining animal entrails." ------------------------------------------------------------------------- Derick Rethans http://derickrethans.nl/ International PHP Magazine http://php-mag.net/ -------------------------------------------------------------------------