A new version of -current has been uploaded to the Web site. Please try
it out and let us know if you run into problems. We consider this to be
1.7RC1.

The main changes since the last -current:

- select() calls have been replaced with calls to poll(). This seems to
be more CPU-efficient and avoids the problems some people ran into when
getting close to the FD_SETSIZE limit.

- the problems with HeadRemove has been fixed - you are now free to
remove any header you want.

- errors resulting from a client closing the socket too early (before
the response was fully sent) are no longer logged.

Of particular interest are issues of portability and performance. poll()
seems to work faster than select(), but there is a question if it is
supported on all systems.
-- 
Robert Segall
Apsis GmbH
Postfach, Uetikon am See, CH-8707
Tel: +41-1-920 4904

Hi Robert --
Are you planning to port to a 2.6 Kernel any time soon?  And to the NPTL?
I am planning our next round of upgrades. 

On Sat, 13 Mar 2004, Robert Segall wrote:

> A new version of -current has been uploaded to the Web site. Please try
> it out and let us know if you run into problems. We consider this to be
> 1.7RC1.
> 
> The main changes since the last -current:
> 
> - select() calls have been replaced with calls to poll(). This seems to
> be more CPU-efficient and avoids the problems some people ran into when
> getting close to the FD_SETSIZE limit.
> 
> - the problems with HeadRemove has been fixed - you are now free to
> remove any header you want.
> 
> - errors resulting from a client closing the socket too early (before
> the response was fully sent) are no longer logged.
> 
> Of particular interest are issues of portability and performance. poll()
> seems to work faster than select(), but there is a question if it is
> supported on all systems.
> -- 
> Robert Segall
> Apsis GmbH
> Postfach, Uetikon am See, CH-8707
> Tel: +41-1-920 4904
> 

Hi,

I am running pound on a 2.6. kernel in a non production enviroment on my 
laptop for serveral weeks now, without any problems.


Dennis Allison said the following on 03/13/04 00:48:
> Hi Robert --
> Are you planning to port to a 2.6 Kernel any time soon?  And to the NPTL?
> I am planning our next round of upgrades. 
> 
> On Sat, 13 Mar 2004, Robert Segall wrote:
> 
> 
>>A new version of -current has been uploaded to the Web site. Please try
>>it out and let us know if you run into problems. We consider this to be
>>1.7RC1.
>>
>>The main changes since the last -current:
>>
>>- select() calls have been replaced with calls to poll(). This seems to
>>be more CPU-efficient and avoids the problems some people ran into when
>>getting close to the FD_SETSIZE limit.
>>
>>- the problems with HeadRemove has been fixed - you are now free to
>>remove any header you want.
>>
>>- errors resulting from a client closing the socket too early (before
>>the response was fully sent) are no longer logged.
>>
>>Of particular interest are issues of portability and performance. poll()
>>seems to work faster than select(), but there is a question if it is
>>supported on all systems.
>>-- 
>>Robert Segall
>>Apsis GmbH
>>Postfach, Uetikon am See, CH-8707
>>Tel: +41-1-920 4904
>>


-- 
Mit freundlichen Grüßen                                Joachim Schmitz
......................................................................
AixtraWare eK ..Joachim Schmitz ..www.aixtraware.de ..t: +49-2464-8851
Hüsgenstr. 33a .....d-52457 Aldenhoven .............f: +49-2464-905163

On Sat, 2004-03-13 at 00:48, Dennis Allison wrote:
> Hi Robert --
> Are you planning to port to a 2.6 Kernel any time soon?  And to the NPTL?
> I am planning our next round of upgrades. 

I'm not aware of any problems with either, excepting the sometimes buggy
RedHat implementation of NPTL in 9.0/EV1.

As an aside - we will wait a short while before going to 2.6 on
production machines here. In my view it is still a bit premature to
upgrade.
-- 
Robert Segall
Apsis GmbH
Postfach, Uetikon am See, CH-8707
Tel: +41-1-920 4904

Hi,

On Sam 13.03.2004 00:23, Robert Segall wrote:
>
>A new version of -current has been uploaded to the Web site. Please try
>it out and let us know if you run into problems. We consider this to be
>1.7RC1.

i have get the -current from the Homepage and see that the Typos still
there.

in 'config.c' on 'SessCOOKIE' the error message 'Session URL' shuld be
'Session Cookie'.

in 'pound.c' the '#ifdef  AEMON' should be '#ifdef  DAEMON'.

Please can you also tell me when does you add the Subpattern Matching?

al ;-)

On Fri, 2004-03-19 at 15:36, Alexander Lazic wrote:
> Hi,
> 
> On Sam 13.03.2004 00:23, Robert Segall wrote:
> >
> >A new version of -current has been uploaded to the Web site. Please try
> >it out and let us know if you run into problems. We consider this to be
> >1.7RC1.
> 
> i have get the -current from the Homepage and see that the Typos still
> there.

Fine - but does it work?

> in 'config.c' on 'SessCOOKIE' the error message 'Session URL' shuld be
> 'Session Cookie'.

Are we talking about the same version? I can't see it - please indicate
a line number.

> in 'pound.c' the '#ifdef  AEMON' should be '#ifdef  DAEMON'.

That's NOT a typo. Please do NOT change that. It is done on purpose, so
that the compile-time flag looks like "gcc ... -DAEMON ...".

> Please can you also tell me when does you add the Subpattern Matching?

Whenever you can come up with a good/viable solution that is not too
complex to implement, reasonably simple to configure and doesn't break
other things.
-- 
Robert Segall
Apsis GmbH
Postfach, Uetikon am See, CH-8707
Tel: +41-1-920 4904

Pound-current works fine on my system - at least as well as the last one
I was running :)  I still have that problem where having RootJail
enabled makes everything take forever to download.  Any ideas on that
one?

Heh, I just figured it out, from the discussion earlier regarding the
2.6 kernel and NPTL, which I'm using.  So I ran pound with
LD_ASSUME_KERNEL=2.4.19 and now there's no noticeable difference in
speed using RootJail.  Does NPTL not play nice with chroot() perhaps? 
I'm using the same workaround with mysql actually.  Apparently it really
isn't ready for prime time :(

Sean

On Fre 19.03.2004 15:58, Robert Segall wrote:
>
>Fine - but does it work?

Sorry i can't test because we i am very busy at work :-(

>Are we talking about the same version? I can't see it - please indicate
>a line number.

Sorry was in the wrong Dir.

>Whenever you can come up with a good/viable solution that is not too
>complex to implement, reasonably simple to configure and doesn't break
>other things.

Have you look into the patch which i have send to the list on 

---
Date: Thu, 22 Jan 2004 22:07:13 +0100
Subject: Typofix and feature add request
---

There are only some few lines of code and an Config example.
Is this to short or unuseable for you then please tell me what not clear
that i can correct it ;-)

al ;-)

On Mon, 2004-03-22 at 09:22, Alexander Lazic wrote:
> >Whenever you can come up with a good/viable solution that is not too
> >complex to implement, reasonably simple to configure and doesn't break
> >other things.
> 
> Have you look into the patch which i have send to the list on 

Yes I have, and answered that your patch is a bit too specific for your
problem. In order to add such functionality it would need to be much
more generic - such as support for arbitrary patterns, support for
URL-based sessions, etc. It should also keep the configuration file
relatively simple.
-- 
Robert Segall
Apsis GmbH
Postfach, Uetikon am See, CH-8707
Tel: +41-1-920 4904

On Fri, 2004-03-19 at 20:01, Seán Gabriel wrote:
> Pound-current works fine on my system - at least as well as the last one
> I was running :)  I still have that problem where having RootJail
> enabled makes everything take forever to download.  Any ideas on that
> one?
> 
> Heh, I just figured it out, from the discussion earlier regarding the
> 2.6 kernel and NPTL, which I'm using.  So I ran pound with
> LD_ASSUME_KERNEL=2.4.19 and now there's no noticeable difference in
> speed using RootJail.  Does NPTL not play nice with chroot() perhaps? 
> I'm using the same workaround with mysql actually.  Apparently it really
> isn't ready for prime time :(

Thanks for the info Sean - this goes into the FAQ.
-- 
Robert Segall
Apsis GmbH
Postfach, Uetikon am See, CH-8707
Tel: +41-1-920 4904

On Mon 22.03.2004 18:25, Robert Segall wrote:
>
>Yes I have, and answered that your patch is a bit too specific for your
>problem. In order to add such functionality it would need to be much
>more generic - such as support for arbitrary patterns, support for
>URL-based sessions, etc. It should also keep the configuration file
>relatively simple.

Oh well i look at this points ;-)

al ;-)