Time for 2.0.51 and 2.1.0

[Sander Striker <st...@apache.org>]

Hi,

I'm going to start a T&R cycle for both 2.0 and 2.1 monday.
Objections?

Sander

Re: Time for 2.0.51 and 2.1.0

[Bill Stoddard <bi...@wstoddard.com>]

Sander Striker wrote:
> Hi,
> 
> I'm going to start a T&R cycle for both 2.0 and 2.1 monday.
> Objections?
> 
> Sander
> 

Got a few 2.0 backports from 2.1 I need to drum up support for but otherwise +1

Bill

Re: Time for 2.0.51 and 2.1.0

[Justin Erenkrantz <ju...@erenkrantz.com>]

--On Thursday, August 26, 2004 7:08 PM +0200 Sander Striker 
<st...@apache.org> wrote:

> I'm going to start a T&R cycle for both 2.0 and 2.1 monday.
> Objections?

Vote early and often for APR 1.0 so that 2.1 can use an official 1.0 release 
of APR.  ;-)  -- justin

Re: Time for 2.0.51 and 2.1.0

["William A. Rowe, Jr." <wr...@rowe-clan.net>]

At 12:59 PM 9/2/2004, Mladen Turk wrote:
>William A. Rowe, Jr. wrote:
>
>>>Could we hope the dev team to relax the situation for mod_proxy/ajp in
>>>future 2.0.x release
>>[...] please don't expect them
>>to sympathize when n.x.z -> n.x.(z+1) starts breaking things, this
>>undermines the confidence in one of the most successful open source
>>projects in the world.
>
>Although I'm not very happy with that, I must say that you hold
>'all the aces' :).

httpd, like Tomcat, is a meritocracy.  Jeff and I argued for a stable
tree and development effort not to break existing 2.0 users.  Others
argued against that.  The last round of voting our suggestion won.
Next vote that could change of course.

Just wanted to reinforce, no one individual or even group of devs
ever holds all the aces here :)

>The major problem is how to prove the new proxy stuff is stable

Aye - there is the rub.

Bill 

Re: Time for 2.0.51 and 2.1.0

[Brian Akins <ba...@web.turner.com>]

On Thu, 2004-09-02 at 13:59, Mladen Turk wrote:

> > You have a simple option though - grab 2.0.  Replace the modules/proxy/
> > tree with 2.1-dev and voila - buildconf - configure - make install.


Any docs on how to actually use the balancer?  Looking at the source
isn't helping me :)

Thanks..

-- 
Brian Akins
Lead Systems Engineer
CNN Internet Operations

Re: Time for 2.0.51 and 2.1.0

["William A. Rowe, Jr." <wr...@rowe-clan.net>]

At 12:59 PM 9/2/2004, Mladen Turk wrote:
>William A. Rowe, Jr. wrote:
>
>>>Could we hope the dev team to relax the situation for mod_proxy/ajp in
>>>future 2.0.x release
>>[...] please don't expect them
>>to sympathize when n.x.z -> n.x.(z+1) starts breaking things, this
>>undermines the confidence in one of the most successful open source
>>projects in the world.
>
>Although I'm not very happy with that, I must say that you hold
>'all the aces' :).

httpd, like Tomcat, is a meritocracy.  Jeff and I argued for a stable
tree and development effort not to break existing 2.0 users.  Others
argued against that.  The last round of voting our suggestion won.
Next vote that could change of course.

Just wanted to reinforce, no one individual or even group of devs
ever holds all the aces here :)

>The major problem is how to prove the new proxy stuff is stable

Aye - there is the rub.

Bill 


---------------------------------------------------------------------
To unsubscribe, e-mail: tomcat-dev-unsubscribe@jakarta.apache.org
For additional commands, e-mail: tomcat-dev-help@jakarta.apache.org

Re: Time for 2.0.51 and 2.1.0

[Mladen Turk <mt...@apache.org>]

William A. Rowe, Jr. wrote:

>>Bad news for me and many others since without AJP support included in
>>2.0.x, users will still require to have mod_jk to link there HTTPD to
>>Tomcats.
>>
>>Could we hope the dev team to relax the situation for mod_proxy/ajp in
>>future 2.0.x release
> 
> Admins understand why n.x -> (n+1).0 can sometimes break things.
> That's the natural aversion to .0 releases.  They can even understand
> if n.x.z -> n.(x+1).0 breaks things.  But please don't expect them
> to sympathize when n.x.z -> n.x.(z+1) starts breaking things, this
> undermines the confidence in one of the most successful open source
> projects in the world.
>

Although I'm not very happy with that, I must say that you hold
'all the aces' :).
The major problem is how to prove the new proxy stuff is stable
enough. Obviously we need a 2.1 release that will
'stick around' for a while, and prove the module's stability.

Further more, all this really looks like a 'killer' 2.1 feature,
with all that protocol and dynamic balancing stuff included.

What I'm concerned about right now is not whether it will be
backported or not, but rather does it works or not. The backport
discussion will be irrelevant when the later gets resolved.

> 
> You have a simple option though - grab 2.0.  Replace the modules/proxy/
> tree with 2.1-dev and voila - buildconf - configure - make install.
> Or use 2.1-dev and help the effort of identifying when 2.1-dev reaches
> release quality.  Mladen provided both alternatives in the current mod.
> 

We can even put some documentation about that to the jakarta site.
Since we have more then enthusiastic user base, I'm sure we'll receive
a lots of valuable real-world test's. But again, we may also encourage
users to use it's 'native' environment (being 2.1).

Regards,
MT.

Re: Time for 2.0.51 and 2.1.0

[Mladen Turk <mt...@apache.org>]

William A. Rowe, Jr. wrote:

>>Bad news for me and many others since without AJP support included in
>>2.0.x, users will still require to have mod_jk to link there HTTPD to
>>Tomcats.
>>
>>Could we hope the dev team to relax the situation for mod_proxy/ajp in
>>future 2.0.x release
> 
> Admins understand why n.x -> (n+1).0 can sometimes break things.
> That's the natural aversion to .0 releases.  They can even understand
> if n.x.z -> n.(x+1).0 breaks things.  But please don't expect them
> to sympathize when n.x.z -> n.x.(z+1) starts breaking things, this
> undermines the confidence in one of the most successful open source
> projects in the world.
>

Although I'm not very happy with that, I must say that you hold
'all the aces' :).
The major problem is how to prove the new proxy stuff is stable
enough. Obviously we need a 2.1 release that will
'stick around' for a while, and prove the module's stability.

Further more, all this really looks like a 'killer' 2.1 feature,
with all that protocol and dynamic balancing stuff included.

What I'm concerned about right now is not whether it will be
backported or not, but rather does it works or not. The backport
discussion will be irrelevant when the later gets resolved.

> 
> You have a simple option though - grab 2.0.  Replace the modules/proxy/
> tree with 2.1-dev and voila - buildconf - configure - make install.
> Or use 2.1-dev and help the effort of identifying when 2.1-dev reaches
> release quality.  Mladen provided both alternatives in the current mod.
> 

We can even put some documentation about that to the jakarta site.
Since we have more then enthusiastic user base, I'm sure we'll receive
a lots of valuable real-world test's. But again, we may also encourage
users to use it's 'native' environment (being 2.1).

Regards,
MT.

Re: Time for 2.0.51 and 2.1.0

["William A. Rowe, Jr." <wr...@rowe-clan.net>]

At 09:25 AM 9/2/2004, Henri Gomez wrote:
>Bad news for me and many others since without AJP support included in
>2.0.x, users will still require to have mod_jk to link there HTTPD to
>Tomcats.
>
>Could we hope the dev team to relax the situation for mod_proxy/ajp in
>future 2.0.x release, since Graham, Mladen and Jean-Frederic works
>hard to make mod_proxy as stable as possible even now with AJP support
>?

Henri, meaning no disrespect to this team (they have accomplished some
fantastic things) ... it took us a very long time to beyond the initial
release to get proxy to something resembling the stability of 1.3, and
mod_cache is still getting there.

Admins understand why n.x -> (n+1).0 can sometimes break things.
That's the natural aversion to .0 releases.  They can even understand
if n.x.z -> n.(x+1).0 breaks things.  But please don't expect them
to sympathize when n.x.z -> n.x.(z+1) starts breaking things, this
undermines the confidence in one of the most successful open source
projects in the world.

When we say x.even releases are -stable- we mean, don't expect things
to break when you upgrade.  When we say x.odd releases are development,
that means ya - it might break on a point bump, live with it and let
us know so the subsequent x.even release is golden.

We are trying to move from the 1.3 and early 2.0 models of change 
this, dabble in that, add this, break that and keep the users 
chasing a moving target.  The beginnings of the 2.0 release were
pretty violent in terms of the changes needed to move from 2.0.36 
to 2.0.39 to 2.0.43, and it didn't do alot for us in terms of
uptake for the httpd-2 server.  Since .43 the consistency has
changed alot - and we want to continue to improve the quality,
and not introduce potentially shaky changes.

You have a simple option though - grab 2.0.  Replace the modules/proxy/
tree with 2.1-dev and voila - buildconf - configure - make install.
Or use 2.1-dev and help the effort of identifying when 2.1-dev reaches
release quality.  Mladen provided both alternatives in the current mod.

Bill

Re: Time for 2.0.51 and 2.1.0

["William A. Rowe, Jr." <wr...@rowe-clan.net>]

At 09:25 AM 9/2/2004, Henri Gomez wrote:
>Bad news for me and many others since without AJP support included in
>2.0.x, users will still require to have mod_jk to link there HTTPD to
>Tomcats.
>
>Could we hope the dev team to relax the situation for mod_proxy/ajp in
>future 2.0.x release, since Graham, Mladen and Jean-Frederic works
>hard to make mod_proxy as stable as possible even now with AJP support
>?

Henri, meaning no disrespect to this team (they have accomplished some
fantastic things) ... it took us a very long time to beyond the initial
release to get proxy to something resembling the stability of 1.3, and
mod_cache is still getting there.

Admins understand why n.x -> (n+1).0 can sometimes break things.
That's the natural aversion to .0 releases.  They can even understand
if n.x.z -> n.(x+1).0 breaks things.  But please don't expect them
to sympathize when n.x.z -> n.x.(z+1) starts breaking things, this
undermines the confidence in one of the most successful open source
projects in the world.

When we say x.even releases are -stable- we mean, don't expect things
to break when you upgrade.  When we say x.odd releases are development,
that means ya - it might break on a point bump, live with it and let
us know so the subsequent x.even release is golden.

We are trying to move from the 1.3 and early 2.0 models of change 
this, dabble in that, add this, break that and keep the users 
chasing a moving target.  The beginnings of the 2.0 release were
pretty violent in terms of the changes needed to move from 2.0.36 
to 2.0.39 to 2.0.43, and it didn't do alot for us in terms of
uptake for the httpd-2 server.  Since .43 the consistency has
changed alot - and we want to continue to improve the quality,
and not introduce potentially shaky changes.

You have a simple option though - grab 2.0.  Replace the modules/proxy/
tree with 2.1-dev and voila - buildconf - configure - make install.
Or use 2.1-dev and help the effort of identifying when 2.1-dev reaches
release quality.  Mladen provided both alternatives in the current mod.

Bill



---------------------------------------------------------------------
To unsubscribe, e-mail: tomcat-dev-unsubscribe@jakarta.apache.org
For additional commands, e-mail: tomcat-dev-help@jakarta.apache.org

Re: Time for 2.0.51 and 2.1.0

[Nick Kew <ni...@webthing.com>]

On Thu, 2 Sep 2004, Henri Gomez wrote:

> Bad news for me and many others since without AJP support included in
> 2.0.x, users will still require to have mod_jk to link there HTTPD to
> Tomcats.
>
> Could we hope the dev team to relax the situation for mod_proxy/ajp in
> future 2.0.x release, since Graham, Mladen and Jean-Frederic works
> hard to make mod_proxy as stable as possible even now with AJP support
> ?

Have you tried running the new proxy code with 2.0.x?  It worked fine
last time I tested it seriously (following my updates to mod_proxy at
the end of June).

That way you're testing the new code in a stable harness.

-- 
Nick Kew

Re: Time for 2.0.51 and 2.1.0

[Henri Gomez <he...@gmail.com>]

Bad news for me and many others since without AJP support included in
2.0.x, users will still require to have mod_jk to link there HTTPD to
Tomcats.

Could we hope the dev team to relax the situation for mod_proxy/ajp in
future 2.0.x release, since Graham, Mladen and Jean-Frederic works
hard to make mod_proxy as stable as possible even now with AJP support
?

Regards

On Thu, 02 Sep 2004 08:59:55 -0500, William A. Rowe, Jr.
<wr...@rowe-clan.net> wrote:
> 
> 
> At 07:55 AM 9/2/2004, Henri Gomez wrote:
> >Should we see the works on mod_proxy and ajp support in the upcoming 2.0.51 ?
> 
> No, not in 2.0.51 (and to your following question of 2.0.x later
> releases, also likely no.)
> 
> Yes, in 2.1.0.
> 
> It was commented that mod_proxy is becoming quite stable - bug
> fixes are being backported but in general, a good number of the
> developers are focusing their efforts at getting the first 2.1.x
> unstable release out the door, and working on mod_cache and other
> issues to get a 2.2.0 general release out the door.
> 
> By releasing 2.1-dev several times, those users interested in
> proxy_ajp and other new proxy features can help shake out the
> issues before this code is branded 2.2.
> 
> The focus for 2.0.x is bug fixes, and 2.2.0 is the next major
> feature release.  It's evens/odds versioning between stable
> and unstable development code.  With the change, the API of
> both 2.0 and 2.2 should be stable for developers.  The changes
> in mod_proxy don't lend themselves to binary compatibility
> with any third party extensions developers had built around
> 2.0 mod_proxy.
> 
> Bill
> 
>

Re: Time for 2.0.51 and 2.1.0

[Henri Gomez <he...@gmail.com>]

Bad news for me and many others since without AJP support included in
2.0.x, users will still require to have mod_jk to link there HTTPD to
Tomcats.

Could we hope the dev team to relax the situation for mod_proxy/ajp in
future 2.0.x release, since Graham, Mladen and Jean-Frederic works
hard to make mod_proxy as stable as possible even now with AJP support
?

Regards

On Thu, 02 Sep 2004 08:59:55 -0500, William A. Rowe, Jr.
<wr...@rowe-clan.net> wrote:
> 
> 
> At 07:55 AM 9/2/2004, Henri Gomez wrote:
> >Should we see the works on mod_proxy and ajp support in the upcoming 2.0.51 ?
> 
> No, not in 2.0.51 (and to your following question of 2.0.x later
> releases, also likely no.)
> 
> Yes, in 2.1.0.
> 
> It was commented that mod_proxy is becoming quite stable - bug
> fixes are being backported but in general, a good number of the
> developers are focusing their efforts at getting the first 2.1.x
> unstable release out the door, and working on mod_cache and other
> issues to get a 2.2.0 general release out the door.
> 
> By releasing 2.1-dev several times, those users interested in
> proxy_ajp and other new proxy features can help shake out the
> issues before this code is branded 2.2.
> 
> The focus for 2.0.x is bug fixes, and 2.2.0 is the next major
> feature release.  It's evens/odds versioning between stable
> and unstable development code.  With the change, the API of
> both 2.0 and 2.2 should be stable for developers.  The changes
> in mod_proxy don't lend themselves to binary compatibility
> with any third party extensions developers had built around
> 2.0 mod_proxy.
> 
> Bill
> 
>

---------------------------------------------------------------------
To unsubscribe, e-mail: tomcat-dev-unsubscribe@jakarta.apache.org
For additional commands, e-mail: tomcat-dev-help@jakarta.apache.org

Re: Time for 2.0.51 and 2.1.0

["William A. Rowe, Jr." <wr...@rowe-clan.net>]

At 07:55 AM 9/2/2004, Henri Gomez wrote:
>Should we see the works on mod_proxy and ajp support in the upcoming 2.0.51 ?

No, not in 2.0.51 (and to your following question of 2.0.x later 
releases, also likely no.)

Yes, in 2.1.0.

It was commented that mod_proxy is becoming quite stable - bug
fixes are being backported but in general, a good number of the
developers are focusing their efforts at getting the first 2.1.x
unstable release out the door, and working on mod_cache and other 
issues to get a 2.2.0 general release out the door.

By releasing 2.1-dev several times, those users interested in
proxy_ajp and other new proxy features can help shake out the
issues before this code is branded 2.2.

The focus for 2.0.x is bug fixes, and 2.2.0 is the next major
feature release.  It's evens/odds versioning between stable
and unstable development code.  With the change, the API of
both 2.0 and 2.2 should be stable for developers.  The changes
in mod_proxy don't lend themselves to binary compatibility
with any third party extensions developers had built around
2.0 mod_proxy.

Bill

Re: Time for 2.0.51 and 2.1.0

[Henri Gomez <he...@gmail.com>]

Should we see the works on mod_proxy and ajp support in the upcoming 2.0.51 ?

Regards

Re: Time for 2.0.51 and 2.1.0

[Joe Orton <jo...@redhat.com>]

On Wed, Sep 01, 2004 at 11:28:31PM +0200, Sander Striker wrote:
> > On Wed, Sep 01, 2004 at 06:32:12PM +0200, Sander Striker wrote:
> > > Something got in the way.  I've got a round tuit reserved for today though ;)
> > 
> > Would be good if you could pick up the tip of the APR/-util 0.9 branches
> > to avoid the build failures on the revisions you tagged.
> 
> ?  I built both tags (prior to tagging) locally.  But yah, if it helps, I can certainly

... build failures for old Linuxes, sorry, I should have qualified that.

Re: Time for 2.0.51 and 2.1.0

[Sander Striker <st...@apache.org>]

----- Original Message ----- 
From: "Joe Orton" <jo...@redhat.com>
To: <de...@httpd.apache.org>
Sent: Wednesday, September 01, 2004 7:27 PM
Subject: Re: Time for 2.0.51 and 2.1.0


> On Wed, Sep 01, 2004 at 06:32:12PM +0200, Sander Striker wrote:
> > Something got in the way.  I've got a round tuit reserved for today though ;)
> 
> Would be good if you could pick up the tip of the APR/-util 0.9 branches
> to avoid the build failures on the revisions you tagged.

?  I built both tags (prior to tagging) locally.  But yah, if it helps, I can certainly
use a new tag of APR.  For the 2.1 line I'm relying on the 1.0 release of APR
(I built locally using 1.0.0-rc6 FWIW).


Sander

Re: Time for 2.0.51 and 2.1.0

[Joe Orton <jo...@redhat.com>]

On Wed, Sep 01, 2004 at 06:32:12PM +0200, Sander Striker wrote:
> Something got in the way.  I've got a round tuit reserved for today though ;)

Would be good if you could pick up the tip of the APR/-util 0.9 branches
to avoid the build failures on the revisions you tagged.

joe

Re: Time for 2.0.51 and 2.1.0

[Sander Striker <st...@apache.org>]

From: "Jess Holle" <je...@ptc.com>
Sent: Wednesday, September 01, 2004 5:04 PM


> Sander Striker wrote:
> 
> >Hi,
> >
> >I'm going to start a T&R cycle for both 2.0 and 2.1 monday.
> >Objections?
> >
> >Sander
> >
> >  
> >
> How is this going?
> 
> [Anxiously awaiting 2.0.51 tarballs...]

Something got in the way.  I've got a round tuit reserved for today though ;)

Sander

Re: Time for 2.0.51 and 2.1.0

[Jess Holle <je...@ptc.com>]

Sander Striker wrote:

>Hi,
>
>I'm going to start a T&R cycle for both 2.0 and 2.1 monday.
>Objections?
>
>Sander
>
>  
>
How is this going?

[Anxiously awaiting 2.0.51 tarballs...]

--
Jess Holle