2007/7/25, Alexei Fedotov <[EMAIL PROTECTED]>:

Hello Sean,

Disclaimer: I'm not a lawyer. :-)

Mauve has a pure GPL license [1]. Executing mauve tests is
establishing a dynamic link between Harmony VM an these tests.


Thanks for your reminding to clear my confusion :)

For me
this means that the linking infrastructure is tainted with GPL and
cannot be a part of BTI.


What i suggest exactly is that end user will build mauve test code locally.
Then tell our bti adaptor the location where the compiled classes are as
a required parameter.
So we will maintain our adaptor script in our svn repos without any GPL
source code.
We just run the test supplied by end user, and we won't:

1. we haven't and won't modify mauve's code.
2. we haven't and won't redistribute mauve's code

So i think it's ok to include mauve test into our BTI through this way, am i
right?
Please correct me if I'm wrong. Thanks.


In other words, if you commit such
infrastructure into Apache repository, the link may help FSF to claim
that Harmony VM links to GPL-ed code and should be re-licensed under
GPL.

All, please correct me if I'm wrong. Thanks.

[1] License,
http://sourceware.org/cgi-bin/cvsweb.cgi/~checkout~/mauve/COPYING?rev=1.1&content-type=text/plain&cvsroot=mauve

On 7/25/07, Sean Qiu <[EMAIL PROTECTED]> wrote:
> Hi, guys
>
> The Mauve Project[1] is collaborative project whose goal is to create a
Free
> suite of functional, black box, tests for the core Java™
> libraries.Theinitial group of its contributors come from the GNU
> Classpath, GCJ, and the
> Kaffe project. All three groups are working on Free (with source)
cleanroom
> implementations of the Java core libraries. So mauve is a clean test
project
> for GNU Classpath to help raise its quality. What's more, its test
> coverage[2] seems pretty good.
>
> The aim of mauve is similar to our unit test, so maybe we can integrate
it
> into our BTI to qualify our harmony's implementation.
> Obviously, it is based on gpl so there are no restrictions on running
the
> testsuite with our own implementation.
>
> But i am not sure we can integrate its source code into our BTI. Its
> home page doesn't supply a binary one yet.
> So maybe we can build the source code into somekind binary one and then
put
> it somewhere for running.
>
> Is this approach OK? Is there anybody interested in this? Any comments?
>
>
> [1]. http://sourceware.org/mauve
> [2]. http://www.object-refinery.com/classpath/mauve/report/
> --
> Sean Qiu
>


--
With best regards,
Alexei,
ESSD, Intel




--
Sean Qiu

Reply via email to