Re: [larm-dev] Jakarta HTTPClient

[Clemens M. <Cle...@in...>]

> Do you think it would be easier and maybe faster if we reported
> your observations to httpclient-dev and suggested the
> alternative approach that you described below?

This could be a viable approach. I would also like a method like
getResponseData(int maxBytes, int timeout) since getResponseData() may a)
produce an indefinitely large byte array and b) may not return at all.

> Also, in which sources are the caches that you mentioned?  JDK
> 1.4.* or HTTPClient?

It's in the sources of java.net.Socket and java.net.InetAddress of 1.4.1_02

> If in HTTPClient, is that in the CVS version or some released
> version?

I used the beta 1 of HTTPClient 2.0 which was released on May 25.

> If this cache stuff is in JDK 1.4.*, then maybe we should see
> what 1.5 brings when it comes out.  I heard that it should be
> out this Fall.  That may be worth waiting.

I would like to use the JDK 1.5 features as soon as possible (enums,
generics).

> Also, with all the LARM things, I think we should try not to get
> stuck with 'details' (this is not a detail in the long run, but
> I think you want to try to put more pieces together before
> thinking about how to improve individual components, tune them,
> etc.).

I know, I change from the broad picture to the details and back. Don't
worry.

> (please don't take this comment as a bad criticism, I'm trying
> to be constructive here :))

No, go ahead. That's why it's important to go public. We all need sparring
partners, that's why we're here. I'm still waiting for your comments on the
docs. If it's easier for you write them into the files themselves (I checked
the changes in now).

Clemens