jsr339-experts@jax-rs-spec.java.net

[jsr339-experts] Re: [jax-rs-spec users] Re: Re: Proposal to downgrade [JAX_RS_SPEC-39] Client Cache Support to MINOR

From: Jan Algermissen <algermissen1971_at_me.com>
Date: Tue, 10 May 2011 19:00:27 +0200

(Sorry for top posting - but it does not really fit inline)

One thing to consider is that there is a significant difference between just any cache and the user agent's private cache. Because cache-control can differentiate between the two.

In my experience, in an enterprise IT context, public caching is not that interesting because you want to avoid the possible defects of whatever public cache is out there in your system. Private caching however is essential for building a performant system then. (REST's performance depends on caching by design)

In my opinion, support for private caching is one of *the* mission critical aspects and Java clients have lacked this ability far to long.

I am with Markus there.

Jan



On May 10, 2011, at 6:40 PM, Markus KARG wrote:

> I doubt that anywhere along the long line between my browser and my favorite
> online shop, the shop software vendor had any chance to influence the cache
> settings applied by me, my provider, or the provider's uplink...
>
>> -----Original Message-----
>> From: Santiago Pericas-Geertsen
>> [mailto:Santiago.PericasGeertsen_at_oracle.com]
>> Sent: Montag, 9. Mai 2011 21:23
>> To: jsr339-experts_at_jax-rs-spec.java.net
>> Subject: [jsr339-experts] Re: [jax-rs-spec users] Re: Re: Proposal to
>> downgrade [JAX_RS_SPEC-39] Client Cache Support to MINOR
>>
>>
>> On May 9, 2011, at 1:08 PM, Markus KARG wrote:
>>
>>>
>>>> Yes, exactly. Possible but non-trivial is an accurate statement. I'm
>>>> also skeptical about a single caching configuration that would work
>>>> well for all use cases out of the box.
>>>
>>> I wonder how the web is working then, since there are lots of caches
>> all the
>>> way and you have no access to their config at all.
>>>
>>
>> Even if I can't access them, it does not imply they aren't configured
>> and tuned by other people based on some requirements. From my point of
>> view, this set of requirements can vary significantly when considering
>> the Client API use cases for a one-size-fits-all, transparent cache to
>> be mandated.
>>
>> -- Santiago
>
>