- ... Commenting over here; I don t want to get an account over there just to add a comment... Anyway: the issue of PUT and DELETE when being repeatedMessage 1 of 70 , Jan 1, 2007View SourceElliotte Harold schrieb:
>Commenting over here; I don't want to get an account over there just to
> ROn 12/2/06, Bill de hOra <bill@dehora. net <mailto:bill%40dehora.net>>
> > >
> > > It often seems to me that POST is a catchall for "do it". I guess the
> > > reason I'm pushing on this is to figure out why the 2 verb web is a
> > > runaway success. What's driving that? Is it because of the design
> > > principles? Or because POST is so decoupled from meaning? Or would
> > > things simply be much better even we had a broader vocabulary deployed
> > > under the design principles?
> I just did an interview with Bill Venners on exactly this:
> http://www.artima com/lejava/ articles/ why_put_and_ delete.html
add a comment...
Anyway: the issue of PUT and DELETE when being repeated potentially
causing overlapping updates/deletes with somebody else's changes is
solved in HTTP. Just send an "If-Match" request header with the last
Etag you got from the server
If you're lucky and the remote server knows how to deal with that
properly, you can even avoid to re-send a large entity with PUT by using
the "Expect: 100-continue" mechanism
With respect to the original topic: can anybody point out why having
*more* verbs would not be REST-ful? For instance, consider PATCH and
Best regards, Julian
- ... As long as we are being brutally honest about how things work in *the real world*, I ll just point out that if some VP discovered that he couldn t edit hisMessage 70 of 70 , Jan 17, 2007View SourceOn 1/9/07, Nic James Ferrier <nferrier@...> wrote:
>As long as we are being brutally honest about how things work
> I think Elliotte is correct that we could fix the problem by getting
> the proxy makers to change their proxies.
> However, how long would it take to fix the problem. The big
> organization that I was referring to had (amongst others) a Novell
> Netware proxy server. It was at least 10 years old.
> I recently made a trip to a medium sized company who were still using
> Microsoft Proxy Server 1.0. I don't even want to think about how old
> that is.
in *the real world*, I'll just point out that if some VP
discovered that he couldn't edit his blog from his shiny new Nokia phone
because the proxy server blocked PUT requests, then that proxy would
get changed so fast it would make your eyes bleed.
To put this in perspective we're talking about a configuration
option on an HTTP proxy/firewall in a company that 11 years
ago was probably running SNA over token ring. To pretend that
things will stay the same as they are today is, at best, delusional.
Joe Gregorio http://bitworking.org