X-HTTP-Method-Override and HTTP Extension Framework

Don Box's Spoutlet

Syndication

I couldn't help but wonder after reading about GData's X-HTTP-Method-Override header whether the Googlies will come to the same conclusion we did.

The initial drafts of SOAP were built on top of the HTTP Extension Framework.

In those early drafts, we required clients to first use M-POST and then only back down to using POST if there were issues getting the M-POST method through the web infrastructure.

You know what? It was a real PITA and we eventually cut it and went with the flow by just using POST (snide comment about ignoring GET goes here :-)).

If I were building a GData client, I honestly wonder why I'd bother using DELETE and PUT methods at all given that X-HTTP-Method-Override is going to work in more cases/deployments.


Posted Jan 16 2007, 01:36 PM by don-box

Comments

Mark Baker wrote re: X-HTTP-Method-Override and HTTP Extension Framework
on 01-16-2007 9:07 PM
I expect that header will be blocked eventually because it exists to subvert firewalls.
Jason Haley wrote Interesting Finds: January 16, 2007
on 01-16-2007 10:06 PM
Jason Haley wrote Interesting Finds: January 16, 2007
on 01-16-2007 10:07 PM
RichB wrote re: X-HTTP-Method-Override and HTTP Extension Framework
on 01-16-2007 10:12 PM
There are GData APIs in C# and Java, so you tend to program against GData sources using these. They attempt standard HTTP verbs before using fallback.

And in any case, there's a small size benefit using the standard verbs. Every byte counts.
RichB wrote re: X-HTTP-Method-Override and HTTP Extension Framework
on 01-16-2007 10:13 PM
Mark Baker wrote:
> I expect that header will be blocked eventually
> because it exists to subvert firewalls.

Firewall blocking of SOAP has proved not to be an issue. So I don't see it being an issue for GData.
Don Box wrote re: X-HTTP-Method-Override and HTTP Extension Framework
on 01-17-2007 1:31 AM
Rich,

Depending on how often PUT and DELETE fail, I can see spending 40 bytes or so to just use POST rather than suffer the extra round trips to discover PUT/DELETE won't work.

I also could swear that some mobile phone stacks don't support PUT/DELETE - but that's pretty old data.

DB
Frank Mantek wrote re: X-HTTP-Method-Override and HTTP Extension Framework
on 01-18-2007 7:37 AM
From what I have been seeing so far, it seems to be the rare case that PUT/DELETE are not supported. We used to have the X-Method-Override be the default in the .NET client, and switched that to off a few months back.

It's in there, as we know of a few firewalls setup to block them, and thought that if the issue is widespread (btw, anyone has any reliable data about this?) it would be better for our customers to have a build in fallback strategy.

So from my point of view, supporting PUT/DELETE is a must to be a more or less Atom Publishing compliant service, which is the what we are trying to be. Supporting this on the client side is hence a given as well.

As most GData clients can probably share a large chunk of code with any other Atom Publishing support code, I don't see the risk that the override will become the default in clients, especially not as we point out that this is a fallback to support non conforming environments.

Regards

Frank Mantek

dbox wrote X-HTTP-Method-Override和Http Extension框架
on 01-31-2007 9:07 AM
??? GData's X-HTTP-Method-Override header?,???????Google???????????????
dbox wrote X-HTTP-Method-Override和Http Extension框架
on 01-31-2007 9:09 AM
??? GData's X-HTTP-Method-Override header?,???????Google???????????????
dbox wrote X-HTTP-Method-Override和Http Extension框架
on 01-31-2007 7:13 PM
??? GData's X-HTTP-Method-Override header?,???????Google???????????????
dbox wrote X-HTTP-Method-Override和Http Extension框架
on 02-01-2007 8:07 PM
??? GData's X-HTTP-Method-Override header?,???????Google???????????????
2girlsonecup com wrote 2girlsonecup com
on 07-09-2008 9:39 AM

Pingback from  2girlsonecup com

Add a Comment

(required)  
(optional)
(required)  
Remember Me?