At 10:42 PM 07/03/2001 -0800, Mo DeJong wrote:
>If I try to install subversion of a fresh Solaris
>box, you can be sure there ain't gonna be a recent version
>of Perl, Python, Tcl, Berkeley DB, or whatever on it.
If I try to install subversion on a fresh Microsoft Windows box, you can be
sure there ain't gonna be a recent version of the Bourne shell.
I think that one has to consider the needs of each population that may be
using subversion, and to keep in mind where running the tests is required
and where it is optional.
1. Developers, who we can just insist have some reasonable set of tools.
These are the ones that will really need to run the test cases the most.
2. Administrators who compile and install a server. As long as they are
using stable releases, the need for them to run the test cases is fairly
low. Sure, it gathers good data on some of the less common platforms if
they do run the tests, but as long as some reasonable subset of
administrators have the set of tools required, there should be sufficient
feedback to deal with most bugs. For administrators that are paranoid
enough to worry about it, they can go and get the tools necessary.
3. Administrators and end users running precompiled binaries for a
particular platform. I wouldn't think that the test cases would even be
included for this group.
The point is, you don't need to run the test cases in order to run
subversion. Unless you are a developer, the tests are completely optional.
For that reason, I would argue very strongly against including the tool
required to run the test cases in subversion. To me, it is identical to a
requirement to include tools for every script and library binding included
with subversion. These are also optional, and that requirement would
clearly be overkill.
Received on Sat Oct 21 14:36:25 2006