> Ok, after a good night and tens of emails this morning on 
> httpclient, I'd
> like to propose a way to go forward, which summarizes what I 
> have understood
> from all these emails :
> 
> Step 1 : Scott Sanders is the release manager for httpclient 
> 1.0. The code
> will be reverted to the one without the new logging stuff. 
> Scott will advise
> us ASAP of the release date and do the release. It needs to 
> be quickly (I
> would say within 1 or 2 weeks max). The 1.0 release should 
> not add any new
> API, not even correct the bugs found and corrected on the 
> rlwrefactoring
> branch (the goal is not to make a perfect release but rather 
> to have one
> that Slide can rely on and if Slide was working with the code 
> as it was
> before, it should continue to work - Moreover correction of the bugs
> introduces API changes). Correction of bugs could be done in 
> version 1.1
> (see step 2).

+1.  My intention to do this would be to have an httpclient that works
in Slide without changing Slide at all.  This is also assuming that the
Slide developers would help to support/maintain the 1.0 release and
branch.  Remy, would you be willing to support this?

> 
> Step 2 : Create a 1.x branch so that bug corrections (but no 
> new features)
> can continue to be made for 1.x versions.

+1

> 
> Step 3 : Once this is done, we move the rlwrefactoring branch 
> to the main
> branch. Remy and other Slide committers should comment on a 
> point by point
> basis on the changes made to that branch (Rod gave a detailed 
> list and this
> can serve for comments). This main branch will be Version 2.0 
> of HttpClient
> as it will introduce several major API changes. We will work 
> with Slide on
> these changes so that Slide can migrate to version 2.0 confidently.

+1

> 
> But please guys, be open for step 3 and let's try to work 
> together in a
> calm, reasoned and professional way. Don't forget we're just 
> coding a stupid
> HttpClient which should have been included in the JDK long 
> ago ! I find it a
> pity that in 2001 we still have to manipulate Socket objects !

:)

Scott Sanders

Reply via email to