Netware port of Subversion 1.0.5 requires testing

[Rodney Crossman <ro...@ravenetcomputers.com>]

Guenter Knauf has posted his port of Subversion 1.0.5 for netware along
with the required patches to build it and bdb 4.2.52  from source.  I
don't know subversion well enough to say that its 100% stable and we are
expecting a few bugs.

I do have 2 working systems up and running on netware 6.5 sp2 with
apache 2.0.50 (which is a current requirement due to apr issues I
believe), and am using auth_ldap for authentication against edirectory. 
I have had some problems with crc errors on commits, though I plan on
testing on both a windows and linux system to make sure its not a
workstation issue.

http://www.gknw.com/development/svn/

I'll add another download location on my own server once I build 1.0.6
when its released.

Rodney



---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: Netware port of Subversion 1.0.5 requires testing

[Mark Benedetto King <mb...@lowlatency.com>]

On Tue, Jul 13, 2004 at 11:11:23PM -0500, Rodney Crossman wrote:
> Guenter Knauf has posted his port of Subversion 1.0.5 for netware along
> with the required patches to build it and bdb 4.2.52  from source.  I
> don't know subversion well enough to say that its 100% stable and we are
> expecting a few bugs.
> 

These patches breaks down into a few categories:

1.) Additional build-system components for netware.

  I think it would be best if these were integrated into the
  configure/gen-make.py system instead of bolted on, as they appear
  to be.  We took great pains to do this for Win32, and I think the
  results have been worth it.

2.) Patches to neon.

  These should be submitted to the neon development team.  They
  look extremely innocuous, so I can't imagine them not being
  accepted.

3.) A patch to subversion/libsvn_fs/fs.c

  This patch uses "#ifdef NETWARE" to determine whether DB_PRIVATE
  should be passed to BDB.  I think it would be more appropriate
  to resolve this at configure time, and generate some sort of
  BDB_NEEDS_DB_PRIVATE #define instead.

  Out of curiosity, are you sure that DB_PRIVATE is necessary
  and correct?  I believe that unless you are using a threaded
  svnserve (and no other repository access method), that
  DB_PRIVATE is not safe to use.


--ben



---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org