Open MPI logo

Open MPI Development Mailing List Archives

  |   Home   |   Support   |   FAQ   |   all Development mailing list

Subject: Re: [OMPI devel] Minor OMPI SVN configuration change
From: Jeff Squyres (jsquyres_at_[hidden])
Date: 2011-02-17 11:11:24


On Feb 17, 2011, at 10:49 AM, Barrett, Brian W wrote:

> Why did "we" make this change? It was originally this way, and we changed it to the no-auth way for a reason.

Are you obliquely saying that there's a reason to have no-auth https access?

DongInn and I changed it because we're still having a problem with svnsync on the OMPI repo:

-----
% svnsync init --username jsquyres file://`pwd`/foomirror https://svn.open-mpi.org/svn/ompi
% svnsync sync file://`pwd`/foomirror
[...MUCH output and about 12 hours...]
Transmitting file data ............................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................svnsync: REPORT of 'https://svn.open-mpi.org/svn/ompi': Could not read chunk size: Secure co
nnection truncated (https://svn.open-mpi.org)
%
-----

I did some googling and (re)discovered a bug report that DongInn and I submitted long ago about svnsync:

    http://subversion.tigris.org/issues/show_bug.cgi?id=3080

The guidance from the SVN developers was to have one method for open+closed (i.e., anon RO access, except for the closed trees) and have a 2nd method for auth-only. I checked with DongInn and we had never followed up on their guidance. I figured we might as well do that as a first step and see if that solved the svnsync problem.

Unfortunately, it didn't. DongInn is still investigating.

-- 
Jeff Squyres
jsquyres_at_[hidden]
For corporate legal information go to:
http://www.cisco.com/web/about/doing_business/legal/cri/