On 30/11/09 19:33, Luke Kanies wrote:
> On Nov 30, 2009, at 8:27 AM, Markus wrote:
>
>>>>
>>>> It could be even simpler, I think.
>>>>
>>>> * Switch to issuing a POST everywhere that we presently issue a GET,
>>>> with a fallback to GET if the POST is rejected (405) or even better
>>>> based on the api version for mixed system backwards compatibility.
>>>>
>>>> * Switch to accepting both POST and get where we now accept only
>>>> GET.
>>>
>>>
>>> I'm comfortable with this, and I like the idea of the server being
>>> agnostic about post/get.
>>>
>>> It only gets us halfway, though - we need a POST to return the
>>> results
>>> of another query entirely (i.e., the POST of the Facts should return
>>> the results of a GET of a Catalog).  How would we do that?
>>
>> I'm envisioning something much simpler.  We presently map requests
>> on to
>> HTTP GETs; instead, we should map them onto HTTP POSTs.  That means
>> we've got unlimited payload size in either direction, but otherwise
>> nothing changes.
>
> Again, that doesn't solve the real problem - I want a Facts post to
> return a Catalog.  How, as the client, would I tell the server that?
>
> Or do we just always return a catalog when someone posts facts?  That
> seems a bit overkill.

Call me stupid, but I don't see why you can't POST the facts to the 
catalog indirector.

Each time you need a catalog you must provide the more current facts you 
can, failure to do so (ie using GET) will use the cached facts from the 
<insert here your favorite cache method including memcache/node mgr or 
whatever>.

-- 
Brice Figureau
My Blog: http://www.masterzen.fr/

--

You received this message because you are subscribed to the Google Groups 
"Puppet Developers" group.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to 
[email protected].
For more options, visit this group at 
http://groups.google.com/group/puppet-dev?hl=en.


Reply via email to