Sigh. I had hoped to avoid this issue (we went through this a year ago 
or so) by maintaining a parallel patch. Given that people actually use 
the patch, it should get tested and bugfixed outside the 1.3 core. Side 
effects from mod_proxy in the 1.3 core will also get exposed this way. 
Thus the integration testing goes on in parallel too.

So, effectively, the patch is "mod_proxy11" at this time. How and if we 
choose to expose it to general 1.3 users is a different issue. Another 
module is one possibility. Going into 1.3.x after sufficient testing is 
another. Supporting the proxy independent of ASF is yet another. We'll 
see how the testing and this or other proxy discussion turns out. 
There's no grand plan for mod_proxy in 1.3; we're just trying to give 
the users the option to use it.

I made the proxy patch because I figure I did 1.3 users a disservice by 
not having the idea to package the proxy changes as a separate patch a 
year earlier. It simply didn't occur to me then, so I took this path to 
make those patches available. The process is open-ended. The users (if 
any 8^) will let us know what the next step should be, since we're an 
open source project.

There's a reason for the recurrence of "user" in the paragraph above.

So, I guess the hackathon gets another item for discussion. We might as 
well talk about the 1.x and 2.x issues together, no?

Chuck

On Monday, March 26, 2001, at 10:49 PM, Bill Stoddard wrote:

>> Chuck,
>>
>>   It's clear from the last 7 trys and 3 partial successes that we need 
>> to
> be
>> very cautious about the 1.3 tree's releases and stability, a ton o' 
>> folk
> are
>> relying on this.  Win32 10 things fixed, 1 thing broke issues don't 
>> bother
> me,
>> but an underpinning like mod_rewrite or mod_proxy do concern me.
>>
>>   Have you considered the advantage in creating mod_proxy11, similar to
> the
>> way we paralleled mod_digest and mod_auth_digest?
>>
>>   I'd be a strong supporter for taking these as a 'fork' - preserving 
>> the
> existing
>> behavior of mod_proxy and 'doing no harm' to that module.
>>
>
> If you had asked me a year ago, I would have agreed. However, it is 
> clear
> that 1.3 will be around for a -long- time and if the HTTP/1.1 support 
> in the
> proxy is not quite right, we fix it and roll another release. It is 
> really
> not that big a deal. I am in favor of commiting the changes.
>
> Bill
>
>
>
>

Reply via email to