RE: Apache BUG: 36495 : ajp_proxy_connect_backend failure

2006-12-21 Thread Benjamin Cuthbert
Thanks Trent i will give this a go.

On the seond co did you not mean to checkout the apr-util/trunk ? or do i 
need to checkout apr twice ?



Trent Nelson [EMAIL PROTECTED] 
12/21/2006 07:51 AM
Please respond to
dev@httpd.apache.org


To
dev@httpd.apache.org
cc

Subject
RE: Apache BUG: 36495  : ajp_proxy_connect_backend failure






 Here's how I build trunk:
 
 $ mkdir -p apr/apr
 $ mkdir apr/apr-util
 $ mkdir -p httpd/trunk
 
 $ cd apr/apr
 $ svn co http://svn.apache.org/repos/asf/apr/apr/trunk
 $ cd ../apr-util
 $ svn co http://svn.apache.org/repos/asf/apr/apr/trunk
 $ cd ../../httpd
 $ svn co http://svn.apache.org/repos/asf/httpd/httpd/trunk

 I'll be happy to push the above into a page under httpd.apache.org/
 dev as here's how to build HEAD for testing if folks think it
 useful. Or I may put it on my blog.

I find that svn externals can provide a useful alternative to the
approach you've mentioned:

F:\scm\svn\svn.onresolve.com\contrib\trunk\httpd\trunksvn pl -v
Properties on '.':
  svn:externals :
src  http://svn.apache.org/repos/asf/httpd/httpd/trunk
src/srclib/apr  http://svn.apache.org/repos/asf/apr/apr/trunk
src/srclib/apr-iconv
http://svn.apache.org/repos/asf/apr/apr-iconv/trunk
src/srclib/apr-util  http://svn.apache.org/repos/asf/apr/apr-util/trunk
src/srclib/zlib  https://svn.onresolve.com/contrib/tags/zlib-1.2.3.1
src/srclib/openssl
https://svn.onresolve.com/contrib/tags/openssl-0.9.8d.1
src/modules/contrib/mod_auth_sspi
https://svn.onresolve.com/contrib/tags/httpd-mod_auth_sspi-1.0.4-2.2.2.1
/src

Tell you what would be nice:
 svn.apache.org/repos/asf/httpd/httpd/views/
 trunk/
 2.0.x/

Where each 'view' is simply an empty directory with the relevant svn
externals set for building that particular branch.  That way, all you'd
need to do to get a buildable trunk is:

svn co http://svn.apache.org/repos/asf/httpd/views/trunk


 Trent.




---

This e-mail may contain confidential and/or privileged information. If you 
are not the intended recipient (or have received this e-mail in error) 
please notify the sender immediately and destroy this e-mail. Any 
unauthorized copying, disclosure or distribution of the material in this 
e-mail is strictly forbidden.

Re: Apache BUG: 36495 : ajp_proxy_connect_backend failure

2006-12-21 Thread William A. Rowe, Jr.
Trent Nelson wrote:
 
 I'll be happy to push the above into a page under httpd.apache.org/
 dev as here's how to build HEAD for testing if folks think it
 useful. Or I may put it on my blog.
 
 I find that svn externals can provide a useful alternative to the
 approach you've mentioned:

Externals are evil.

They are *very* prone to mis-tagging.  A tag with externals pointing
back to another leaf of a trunk are WORTHLESS.  Several projects have
been guilty of this.

Also we don't require you to grab the latest apr... httpd trunk (so far)
builds with apr 1.2.x.  Someday it will require 1.3.x but that will be
after at least 1.3.0 has shipped and can be installed.

Which means - grab apr trunk *if you want*.

I keep apr-0.9/1.2/1.x trunk checked out (ditto -util/-iconv) and just
symlink from within httpd-2.0/2.2/2.x trunk.  Since checkouts into the
subdir don't update very well, I just svn up httpd-* apr-* from my root.
It makes it much easier to move patches from version to version of our
'srclib's.





Re: Apache BUG: 36495 : ajp_proxy_connect_backend failure

2006-12-21 Thread William A. Rowe, Jr.
Trent Nelson wrote:
 
 Tell you what would be nice:
   svn.apache.org/repos/asf/httpd/httpd/views/

AHHH - ok, retracting my statement... your approach is much less prone
to mis-tagging.  However, it can still yield some pretty obtuse commits
if one is not careful :)


Re: Apache BUG: 36495 : ajp_proxy_connect_backend failure

2006-12-21 Thread Sander Temme


On Dec 21, 2006, at 12:37 AM, William A. Rowe, Jr. wrote:


Trent Nelson wrote:


Tell you what would be nice:
svn.apache.org/repos/asf/httpd/httpd/views/


AHHH - ok, retracting my statement... your approach is much less prone
to mis-tagging.  However, it can still yield some pretty obtuse  
commits

if one is not careful :)


That's simple: don't make that view writable. It'd be a read-only  
service for folks that want to check out something buildable.


It'll still make you configure and build apr + apr-util every time  
you make a change to httpd. Which is my main motivation to keep them  
out of my tree.


S.

--
[EMAIL PROTECTED]http://www.temme.net/sander/
PGP FP: 51B4 8727 466A 0BC3 69F4  B7B8 B2BE BC40 1529 24AF




smime.p7s
Description: S/MIME cryptographic signature


RE: Apache BUG: 36495 : ajp_proxy_connect_backend failure

2006-12-21 Thread Trent Nelson

  Tell you what would be nice:
 svn.apache.org/repos/asf/httpd/httpd/views/
 
  AHHH - ok, retracting my statement... your approach is much less
prone
  to mis-tagging.  However, it can still yield some pretty obtuse
  commits
  if one is not careful :)
 
 That's simple: don't make that view writable. It'd be a read-only
 service for folks that want to check out something buildable.

Indeed; it provides a nice easy way for people wanting to do local
builds, not necessarily developers doing hardcore development (who would
have the savvy to set up the development environment any which way they
want).

Any chance of trialing this view approach for trunk?  Wouldn't take more
than a couple of minutes to set up...

Trent.


RE: Apache BUG: 36495 : ajp_proxy_connect_backend failure

2006-12-21 Thread Trent Nelson
  I find that svn externals can provide a useful alternative to the
  approach you've mentioned:
 
 Externals are evil.
 
 They are *very* prone to mis-tagging.  A tag with externals pointing
 back to another leaf of a trunk are WORTHLESS.  Several projects have
 been guilty of this.

I'd certainly agree that *tagging* a structure that contains externals
pointing to other trunk branches is a recipe for disaster.  As long as
you don't do this, they can be very useful, even if your entire
externals definition is composed of tagged/read-only branches:

http://svn.apache.org/repos/asf/httpd/httpd/views/2.2.3
svn:externals:
src/ http://svn.apache.org/repos/asf/httpd/httpd/tags/2.2.3
src/srclib/apr http://svn.apache.org/repos/asf/apr/apr/tags/1.2.8
src/srclib/apr-util
http://svn.apache.org/repos/asf/apr/apr-util/tags/1.2.8
src/srclib/apr-iconv
http://svn.apache.org/repos/asf/apr/apr-iconv/tags/1.1.1

Thus, I could check out the exact source used to build 2.2.3 via:

svn co http://svn.apache.org/repos/asf/httpd/httpd/views/2.2.3

I'd personally prefer this sort of an approach to, say, relying on the
source tars that are manually rolled at release time.

Trent.



Re: svn commit: r488780 - /httpd/httpd/branches/2.2.x/STATUS

2006-12-21 Thread William A. Rowe, Jr.
Roy T. Fielding wrote:

 So, I'm -0.9 on the patch, even though it isn't much worse than
 the existing code.  I'd be -1 if I could be sure what it did.

And I wouldn't debate it... back to the drawing board.  As an improvement
to the behavior I support it, in terms of style it got a -0.5 from me, but
I didn't have the cycles to clean it up.

Thanks for the kick in the ass to do so.  It's time for the less-than-half-assed
solution to WaitForMultipleObjects.

Bill