Re: HTTP_Client or Net_Socket bug?
| From: | Lukas Smith | Date: | Tue, 16 Nov 2004 22:47:53 +0000 |
| Subject: | Re: HTTP_Client or Net_Socket bug? | ||
| References: | 1 2 3 4 5 | Groups: | php.pear.dev |
| Request: | Send a blank email to pear-dev+get-34421@lists.php.net to get a copy of this message | ||
Greg Beaver wrote:
Chuck Hagenbuch wrote:Essentially the current PEAR works around the memory management issues. A destructor is only registered for those packages that define a destructor. Also the shutdown method is only registered once, at the time PEAR finds the first destructor. So the performance impact should become really low. Especially if you are using several PEAR packages and once you return the first error object the impact is fairly limited. Also expectError() is quite a nice feature to be able to override the global error handler. Although its black magic for most people :-( So to conclude there are less reasons to get into the fuss of breaking BC over not extending PEAR anymore. regards, LukasDo I remember correctly that progress has been made on the impact of extending PEAR? Or only on including it?Both. You should try Net_Socket that extends PEAR with PEAR from pre-1.3.3 and then compare it to the one with PEAR from 1.3.3, and then compare that to the non-pear version, if you want to get a sense of how it has improved.