Content-type bug (http://framework.zend.com/issues/browse/ZF-1615) fixed in r. 5439. Thanks for the report, Federico!
BTW in the future, please try not to submit to fw-general - this kind of issue would be better noticed in [EMAIL PROTECTED] Best regards, Shahar. On Sun, 2007-06-24 at 15:37 +0300, Shahar Evron wrote: > Hi Federico, > > On Sat, 2007-06-23 at 19:31 +0200, Federico Galassi wrote: > > On 23/giu/07, at 18:33:41, Matthew Weier O'Phinney wrote: > > > > > -- Federico Galassi <[EMAIL PROTECTED]> wrote > > > (on Saturday, 23 June 2007, 03:49 PM +0200): > > >> Hello, > > >> i started to write this e-mail as a bug report. For the record, the > > >> bug is: > > >> --- code --- > > >> $uri = "http://www.google.com/";; > > >> $client = new Zend_Http_Client(); > > >> $client->setUri($uri); > > >> $client->request('POST'); > > >> echo $client->getLastRequest(); > > >> $client->request('GET'); > > >> echo $client->getLastRequest(); > > > > > > You need to call $client->resetParameters() between subsequent > > > requests. > > > This is clearly indicated in the manual. > > > > Unfortunately, resetParameters() doesn't reset the content type, but > > that would just make it a simple bug. > > I will report and fix this simple bug ASAP. > > > Does it make sense to leave the component > > in a broken state unless an action is taken by the user, even if it's > > stated by > > documentation? I mean, if it has to be done, why isn't it done implicitly > > after > > every request is made? > > > > Yes. In my opinion, and as it seems in the opinion of everyone who has > been using Zend_Http_Client in the last 6 months or so, this is > perfectly Ok. > > There is always a trade-off between performance + flexibility, and doing > things "magically" in the background for the user. Since we are writing > a framework, and not an end-user application, I tend to go with > performance + flexibility. I don't think that asking the programmer to > call $client->resetParameters() is too much - especially because in 80% > of the cases it makes sense. > > In most cases, one script would use a single $client object to send > requests to the same server / application, using the same client-wide > settings more or less (eg. Accept headers), and usually if > authentication header was set, it would be used on all requests. On the > other hand, if one wants to avoid this, they can always use > $client->resetParameters(). > > If I follow your suggestion, most people who do consecutive requests > will need to reset their client before every request (eg. call several > functions). I don't really like that idea. > > There is sense in decoupling the part that builds and sends HTTP > messages from the client - this way we could have a Zend_Http_Request > object that can be created separately and sent over HTTP, or perhaps > even resend Zend_Controller_Request_Http objects with one or two lines > of code. Those who need to send consecutive requests with similar > features (and do things like capture cookies and follow redirections) > could use Zend_Http_Client as is. This kind of setup makes sense - but > it's not going to happen in the near future, and is a major change - > thus it needs to go through a full proposal cycle. > > Best regards, > > Shahar. > > > > > Thank you, > > Federico > >
signature.asc
Description: This is a digitally signed message part
