Re: svn commit: r180333 - /httpd/httpd/branches/ssl-fips-dev

2005-06-07 Thread Justin Erenkrantz
--On Monday, June 6, 2005 8:24 PM +0200 Sander Striker [EMAIL PROTECTED] 
wrote:



Personally I prefer entire tree branches.


Agreed.  This setup is going to make it difficult to test.  -- justin


Re: svn commit: r180333 - /httpd/httpd/branches/ssl-fips-dev

2005-06-06 Thread André Malo
* [EMAIL PROTECTED] wrote:

 Author: wrowe
 Date: Mon Jun  6 09:22:16 2005
 New Revision: 180333

 URL: http://svn.apache.org/viewcvs?rev=180333view=rev
 Log:
 Sandbox of httpd/trunk/modules/ssl/ for OpenSSL 0.9.7 fips integration
 development

 Added:
 httpd/httpd/branches/ssl-fips-dev/
   - copied from r180332, httpd/httpd/trunk/modules/ssl/

I'm not sure about any policy, but it seems better to me to branch off the 
whole trunk, not just a subtree (you never know, what needs to be modified 
finally)

... Opinions?

nd
-- 
Gefunden auf einer Webdesigner-Seite:
 Programmierung in HTML, XML, WML, CGI, FLASH 

# Andr Malo # http://pub.perlig.de/ #


Re: svn commit: r180333 - /httpd/httpd/branches/ssl-fips-dev

2005-06-06 Thread Paul Querna

Andr Malo wrote:


* [EMAIL PROTECTED] wrote:

 


Author: wrowe
Date: Mon Jun  6 09:22:16 2005
New Revision: 180333

URL: http://svn.apache.org/viewcvs?rev=180333view=rev
Log:
Sandbox of httpd/trunk/modules/ssl/ for OpenSSL 0.9.7 fips integration
development

Added:
   httpd/httpd/branches/ssl-fips-dev/
 - copied from r180332, httpd/httpd/trunk/modules/ssl/
   



I'm not sure about any policy, but it seems better to me to branch off the 
whole trunk, not just a subtree (you never know, what needs to be modified 
finally)


... Opinions?
 


I agree with nd.

-Paul


Re: svn commit: r180333 - /httpd/httpd/branches/ssl-fips-dev

2005-06-06 Thread Garrett Rooney

Paul Querna wrote:

Andr Malo wrote:


* [EMAIL PROTECTED] wrote:

 


Author: wrowe
Date: Mon Jun  6 09:22:16 2005
New Revision: 180333

URL: http://svn.apache.org/viewcvs?rev=180333view=rev
Log:
Sandbox of httpd/trunk/modules/ssl/ for OpenSSL 0.9.7 fips integration
development

Added:
   httpd/httpd/branches/ssl-fips-dev/
 - copied from r180332, httpd/httpd/trunk/modules/ssl/
  



I'm not sure about any policy, but it seems better to me to branch off 
the whole trunk, not just a subtree (you never know, what needs to be 
modified finally)


... Opinions?
 


I agree with nd.


Branching the whole tree is what's normally done.

If it turns out you're only actually using a small subset you can switch 
just that portion of your trunk checkout to the branch, and thus avoid 
the need to continuously merge all changes into the branch in order to 
stay up to date, but it's way easier later on if you decide you need to 
modify something outside that directory if you already have it there on 
the branch.


-garrett


Re: svn commit: r180333 - /httpd/httpd/branches/ssl-fips-dev

2005-06-06 Thread William A. Rowe, Jr.
At 11:43 AM 6/6/2005, Paul Querna wrote:
André Malo wrote:

Sandbox of httpd/trunk/modules/ssl/ for OpenSSL 0.9.7 fips integration
development

Added:
   httpd/httpd/branches/ssl-fips-dev/
 - copied from r180332, httpd/httpd/trunk/modules/ssl/

I'm not sure about any policy, but it seems better to me to branch off the 
whole trunk, not just a subtree (you never know, what needs to be modified 
finally)
I agree with nd.

My thinking was; 98% of the changes will be to modules/ssl code
itself.  90% of the remainder are likely to be incidental bug
fixes than Ben, I or anyone else working on the tree encounter.
The other 10% is autoconf detection.

Most autoconf can be localized in modules/ssl, but for the rest
of the autoconf issues, they should be no-ops if this code is
not adopted.  E.g. detecting if libcrypto.so.fips (the signature
hash file) exists.

Contrawise, most bug fixes to head will be better reviewed if 
those of us working in this branch pick them up immediately.
But if concensus says make this a full tree, I'll be happy to
oblige.

Because most users in crypto-restricted environments would simply
rm -rf modules/ssl  ---  it's VERY important that all our crypto 
code resides in that tree.  That's the basis for my choice.

Bill




Re: svn commit: r180333 - /httpd/httpd/branches/ssl-fips-dev

2005-06-06 Thread Sander Striker

William A. Rowe, Jr. wrote:

At 11:43 AM 6/6/2005, Paul Querna wrote:


André Malo wrote:



Sandbox of httpd/trunk/modules/ssl/ for OpenSSL 0.9.7 fips integration
development

Added:
 httpd/httpd/branches/ssl-fips-dev/
   - copied from r180332, httpd/httpd/trunk/modules/ssl/


I'm not sure about any policy, but it seems better to me to branch off the 
whole trunk, not just a subtree (you never know, what needs to be modified 
finally)


I agree with nd.



My thinking was; 98% of the changes will be to modules/ssl code
itself.  90% of the remainder are likely to be incidental bug
fixes than Ben, I or anyone else working on the tree encounter.
The other 10% is autoconf detection.

Most autoconf can be localized in modules/ssl, but for the rest
of the autoconf issues, they should be no-ops if this code is
not adopted.  E.g. detecting if libcrypto.so.fips (the signature
hash file) exists.

Contrawise, most bug fixes to head will be better reviewed if 
those of us working in this branch pick them up immediately.

But if concensus says make this a full tree, I'll be happy to
oblige.


Personally I prefer entire tree branches.


Sander