[flexcoders] Re: HttpService post error 2032:iostream error

2008-04-08 Thread mailamannow
Hi Dmitri

Can you please explain me if WireShark could prove to be a useful tool
for debugging Web Services. I tried it, but couldn't get it working to
debug my Web Service





[flexcoders] Re: HttpService post error 2032:iostream error

2008-04-07 Thread Dmitri Girski
Install Wireshark sniffer and check the contents of the packets. I am
inclined to think that this problem is on a server side.

PS What is on the server side?



--- In flexcoders@yahoogroups.com, longxjyh [EMAIL PROTECTED] wrote:

 When I use HttpService to send data by post method,I encounter a 
 problem.I find if the length of the post data is over 1022,then flex 
 rises the runtime error 2032.However the data is sended to server 
 wholelly and the server has accepted all the data without any error.I 
 can't understand why flex rises such error.The problem exists only in 
 IE.In firefox ,it's ok,no problem.





[flexcoders] Re: HttpService post error 2032:iostream error

2008-04-07 Thread Carlos Reyes
I believe I've run into the same problem. I haven't checked the length
of the POST data, but I'll bet it's over 1022 bytes.

This is under Windows XP, Internet Explorer, and the latest Flash
(9,0,115,0). I'm seeing the same behavior with Opera under Windows.
Firefox is fine, Apple Safari (under Windows) is fine too. Firefox under
Linux is fine. The data gets to the server okay. It seems that the
problem is related to the Flash ActiveX wrapper.

I'm communicating with Amazon S3, so I don't have a lot of control over
the settings. If I got more information with the error, I could ignore
it. But all I get is 2032 Stream Error. Very frustrating and scary for
my customers (over 50% of which use IE).

I haven't tried WireShark under Windows, but I'm afraid I just don't
have much control over what Amazon S3 responds with.

Carlos


--- In flexcoders@yahoogroups.com, Dmitri Girski [EMAIL PROTECTED] wrote:

 Install Wireshark sniffer and check the contents of the packets. I am
 inclined to think that this problem is on a server side.

 PS What is on the server side?



 --- In flexcoders@yahoogroups.com, longxjyh longxjyh@ wrote:
 
  When I use HttpService to send data by post method,I encounter a
  problem.I find if the length of the post data is over 1022,then flex
  rises the runtime error 2032.However the data is sended to server
  wholelly and the server has accepted all the data without any
error.I
  can't understand why flex rises such error.The problem exists only
in
  IE.In firefox ,it's ok,no problem.
 





[flexcoders] Re: HttpService post error 2032:iostream error

2008-04-07 Thread Carlos Reyes
I checked the packets with Charles (www.xk72.com/charles) under Windows.
The request body is over 1KB.

The response packet just says 204 No Content. There is no body.
Obviously, the ActiveX wrapper is taking this to mean that something
went wrong. Again, since I have no information other than Error 2032,
I don't see how I can safely ignore these errors.

Carlos


--- In flexcoders@yahoogroups.com, Dmitri Girski [EMAIL PROTECTED] wrote:

 Install Wireshark sniffer and check the contents of the packets. I am
 inclined to think that this problem is on a server side.

 PS What is on the server side?



 --- In flexcoders@yahoogroups.com, longxjyh longxjyh@ wrote:
 
  When I use HttpService to send data by post method,I encounter a
  problem.I find if the length of the post data is over 1022,then flex
  rises the runtime error 2032.However the data is sended to server
  wholelly and the server has accepted all the data without any
error.I
  can't understand why flex rises such error.The problem exists only
in
  IE.In firefox ,it's ok,no problem.
 





[flexcoders] Re: HttpService post error 2032:iostream error

2008-04-07 Thread Dmitri Girski
OK, guys, I'll try to summarise my painful experience with HTTPService.

Things which could go wrong:
- HTTPService settings, i.e. type of the parameters object. You should
check that if you send XML, the type of the object is XML and
content-type is application/xml
- I had a lot of mysterious problems with #2032 error, so I ended up
with a solution to create HTTPService object, send data and then
disconnect() it and kill (assign a null). As soon as the FlashPlayer
code is closed unfortunately we have no idea how it works with sockets
and what are the real error codes we get.

-requests could be cached by the browser/server, to prevent this it is
a good idea always add some random number parameter to the request URL
(someURL?nocache=xyz)

-server side (PHP)
magic sequence to prevent caching
header('Pragma: bogus');
header('Cache-control: bogus');
similar things should be done for other servers, check Judah's blog:
http://www.judahfrangipane.com/blog/?p=87

header('Expires: 0') - causes a lot of #2032's

- error #2032 could be produced in certain cases when server replies
with unexpected response. Usually we see #1090 - XML parser error, but
sometimes, #2032 is thrown. I happens when server replies with
anythinh else rather then 200 code and also some problems with PHP -
it is always wise to check server logs for any errors.


-combination Apache2 + SSL + Gentoo + HTTPService does not work. This
is just a matter of fact. I didn't have time nor resources to
investigate it further, but FreeBSD + Apache1 + SSL + HTPPService
works seamlessly. I suspect that Apache2+SSL + FlashPlayer have some
problem with packet resends during the error recovery - this is the
only result I've got looking through megs of Wireshark logfiles. 

The result of this combination was #2032 error due to the timeouts in
the network - if SSL connection is dropped there was a little chance
that it will be renegotiated again.



PS I have not dealt with Charles, but I highly recommend Wireshark -
this thing sits at the lowest level and allows you to see the raw
network packets. And it is free.


Cheers,
Dmitri.




[flexcoders] Re: HttpService post error 2032:iostream error

2008-04-07 Thread Carlos Reyes
I found a workaround. I can tell Amazon S3 that I want to do a redirect.
Now the response code is 303 See Other, which does not generate the
error.

Opera under Windows suffers from the same problem (not Firefox),
although it apparently does not use the ActiveX adaptor. It uses the
PlugIn. I don't explain them, just report them.

Carlos


--- In flexcoders@yahoogroups.com, Dmitri Girski [EMAIL PROTECTED] wrote:

 OK, guys, I'll try to summarise my painful experience with
HTTPService.

 Things which could go wrong:
 - HTTPService settings, i.e. type of the parameters object. You should
 check that if you send XML, the type of the object is XML and
 content-type is application/xml
 - I had a lot of mysterious problems with #2032 error, so I ended up
 with a solution to create HTTPService object, send data and then
 disconnect() it and kill (assign a null). As soon as the FlashPlayer
 code is closed unfortunately we have no idea how it works with sockets
 and what are the real error codes we get.

 -requests could be cached by the browser/server, to prevent this it is
 a good idea always add some random number parameter to the request URL
 (someURL?nocache=xyz)

 -server side (PHP)
 magic sequence to prevent caching
 header('Pragma: bogus');
 header('Cache-control: bogus');
 similar things should be done for other servers, check Judah's blog:
 http://www.judahfrangipane.com/blog/?p=87

 header('Expires: 0') - causes a lot of #2032's

 - error #2032 could be produced in certain cases when server replies
 with unexpected response. Usually we see #1090 - XML parser error, but
 sometimes, #2032 is thrown. I happens when server replies with
 anythinh else rather then 200 code and also some problems with PHP -
 it is always wise to check server logs for any errors.


 -combination Apache2 + SSL + Gentoo + HTTPService does not work. This
 is just a matter of fact. I didn't have time nor resources to
 investigate it further, but FreeBSD + Apache1 + SSL + HTPPService
 works seamlessly. I suspect that Apache2+SSL + FlashPlayer have some
 problem with packet resends during the error recovery - this is the
 only result I've got looking through megs of Wireshark logfiles.

 The result of this combination was #2032 error due to the timeouts in
 the network - if SSL connection is dropped there was a little chance
 that it will be renegotiated again.



 PS I have not dealt with Charles, but I highly recommend Wireshark -
 this thing sits at the lowest level and allows you to see the raw
 network packets. And it is free.


 Cheers,
 Dmitri.





[flexcoders] Re: HttpService post error 2032:iostream error

2008-04-07 Thread Carlos Reyes
Thank you for your reply. I've done a lot of research on this problem.
None of the usual causes/solutions to 2032 errors appear to apply.

This is an HTTP POST without using SSL. It's not a caching problem.
Caching is an issue with GET commands, I know that the request really
goes out, and the body changes with every post. I know that the content
type is fine.

Clearly, it's the 204 No Content return status that is messing things
up. Unless I can get Amazon to return something different, I'm stuck. In
the meantime, I see no choice but to silently throw away these errors.
Horrible, I know, but I have no choice.

I've used WireShark under Linux. Charles has a nicer interface and gives
the same information that I'm looking for.

Carlos


--- In flexcoders@yahoogroups.com, Dmitri Girski [EMAIL PROTECTED] wrote:

 OK, guys, I'll try to summarise my painful experience with
HTTPService.

 Things which could go wrong:
 - HTTPService settings, i.e. type of the parameters object. You should
 check that if you send XML, the type of the object is XML and
 content-type is application/xml
 - I had a lot of mysterious problems with #2032 error, so I ended up
 with a solution to create HTTPService object, send data and then
 disconnect() it and kill (assign a null). As soon as the FlashPlayer
 code is closed unfortunately we have no idea how it works with sockets
 and what are the real error codes we get.

 -requests could be cached by the browser/server, to prevent this it is
 a good idea always add some random number parameter to the request URL
 (someURL?nocache=xyz)

 -server side (PHP)
 magic sequence to prevent caching
 header('Pragma: bogus');
 header('Cache-control: bogus');
 similar things should be done for other servers, check Judah's blog:
 http://www.judahfrangipane.com/blog/?p=87

 header('Expires: 0') - causes a lot of #2032's

 - error #2032 could be produced in certain cases when server replies
 with unexpected response. Usually we see #1090 - XML parser error, but
 sometimes, #2032 is thrown. I happens when server replies with
 anythinh else rather then 200 code and also some problems with PHP -
 it is always wise to check server logs for any errors.


 -combination Apache2 + SSL + Gentoo + HTTPService does not work. This
 is just a matter of fact. I didn't have time nor resources to
 investigate it further, but FreeBSD + Apache1 + SSL + HTPPService
 works seamlessly. I suspect that Apache2+SSL + FlashPlayer have some
 problem with packet resends during the error recovery - this is the
 only result I've got looking through megs of Wireshark logfiles.

 The result of this combination was #2032 error due to the timeouts in
 the network - if SSL connection is dropped there was a little chance
 that it will be renegotiated again.



 PS I have not dealt with Charles, but I highly recommend Wireshark -
 this thing sits at the lowest level and allows you to see the raw
 network packets. And it is free.


 Cheers,
 Dmitri.