Re: What do we get with PUT/DELETE?

[Robert Sayre <mint@xxxxxxxxxxxxxxx>]

On 2/25/04 10:06 PM, "Tim Bray" <tbray@xxxxxxxxxxxxxx> wrote:

> On Feb 24, 2004, at 9:18 PM, Ken MacLeod wrote:
>> 
>> Off the top of my head, we get:
> 
> In your list, I do not observe any entries that amount to "this will
> make it easier to implement" or "this will enable features that would
> not otherwise be possible."  Thus you are missing the two most powerful
> engineering arguments are are facing an up-hill struggle.  In fact, it
> is observable that insisting on PUT & DELETE in fact adds to the
> difficulty of implementation in some contexts; which seems fatal to me
> when the arguments in favor are so philosophical. -Tim
> 

Assuming we are weighing the difference between the alternatives presented
on DifferentlyAbledClients, here is my list:

* Bandwidth & Processor conservation. On a site hosting hundreds or
thousands of blogs, the difference would be significant. Why must Atom
_require_ wrapping http status codes in XML elements? Shades of rfc3252.
Note that this bandwidth increase also hurts the very same client devices
POST-only is supposed to help.

* On client platforms that support PUT/DELETE, insisting on POST makes
implementation more difficult than it otherwise would be.

* It makes the Atom traffic more opaque to network hardware.


Robert Sayre