Tests and copyright

Tests and copyright

[David Engster]

I'm currently migrating our EIEIO test suite to Emacs, and I'm wondering
if files in the tests/ directory fall under the same rules w.r.t. to
copyright/papers?

-David

Re: Tests and copyright

[PJ Weisberg]

[-- Attachment #1: Type: text/plain, Size: 351 bytes --]

On Aug 13, 2013 12:50 PM, "David Engster" <deng@randomsample.de> wrote:
>
> I'm currently migrating our EIEIO test suite to Emacs, and I'm wondering
> if files in the tests/ directory fall under the same rules w.r.t. to
> copyright/papers?

I don't see why they wouldn't,  if they're distributed.  They're just as
much the original work of an author.

[-- Attachment #2: Type: text/html, Size: 501 bytes --]

Re: Tests and copyright

[David Engster]

PJ Weisberg writes:
> On Aug 13, 2013 12:50 PM, "David Engster" <deng@randomsample.de> wrote:
>>
>> I'm currently migrating our EIEIO test suite to Emacs, and I'm wondering
>> if files in the tests/ directory fall under the same rules w.r.t. to
>> copyright/papers?
>
> I don't see why they wouldn't,  if they're distributed.  They're just as much
> the original work of an author.

AFAIK, they are not distributed with the tarballs. Also, at least I do
not consider them part of "the Emacs source code", but I'm not a lawyer.

Anyway, I was just hoping that I wouldn't have to wade through years of
log history, as if porting test code wasn't boring enough already.

-David

Re: Tests and copyright

[Stefan Monnier]

>>> I'm currently migrating our EIEIO test suite to Emacs, and I'm wondering
>>> if files in the tests/ directory fall under the same rules w.r.t. to
>>> copyright/papers?
>> I don't see why they wouldn't,  if they're distributed.  They're just
>> as much the original work of an author.
> AFAIK, they are not distributed with the tarballs.  Also, at least I do
> not consider them part of "the Emacs source code", but I'm not a lawyer.

Actually, this question has popped up in the past, and until now we
haven't had any reason to find a good answer, so we haven't dug
any deeper.

There are sound reasons to consider that the test suite could follow
different guidelines in this respect:
- in case of a legal challenge, we could drop the relevant tests without
  suffering any direct consequence.
- test cases may come simply from bug-reports from many people who
  aren't likely to contribute any further.

My opinion is that it is not necessary to get copyright assignments for
the test cases themselves (e.g. the files in emacs/test/indent, or the
compile-tests--test-regexps-data var in test/automated/compile-tests.el).

That still leaves open the question for the code written to actually run
the tests (e.g. the functions compile--test-error-line and
compile-test-error-regexps in that same
test/automated/compile-tests.el).  To the extent that this code is
likely to have parts that can be reused between different sets of tests
for different packages, I tend to think that it falls in the camp of
"it's just code like any other".

> Anyway, I was just hoping that I wouldn't have to wade through years of
> log history, as if porting test code wasn't boring enough already.

Is the above rule "lax enough" to significantly lighten your workload?


        Stefan

Re: Tests and copyright

[Stephen J. Turnbull]

David Engster writes:

 > > I don't see why they wouldn't [require assignment], if they're
 > > distributed.  They're just as much the original work of an author.
 > 
 > AFAIK, they are not distributed with the tarballs.

Any "conveying" to a third party is distribution for legal purposes.
That's defined in the GPL, and includes publicly accessible VCS repos.

However, there's no law that requires assignment; it's purely a policy
of the GNU Project (at least, I don't think Stefan would be allowed to
unilaterally end it, even if he wanted to) with respect to "GNU
Emacs".  I'm pretty sure that the Emacs maintainers would want to
consider the test suite "part of Emacs", so the same rules would
likely apply.  But you'd have to ask whoever sets such policies (I
would guess RMS, but maybe the FSF Board or Stefan).

Re: Tests and copyright

[David Engster]

Stefan Monnier writes:
> My opinion is that it is not necessary to get copyright assignments for
> the test cases themselves (e.g. the files in emacs/test/indent, or the
> compile-tests--test-regexps-data var in test/automated/compile-tests.el).
>
> That still leaves open the question for the code written to actually run
> the tests (e.g. the functions compile--test-error-line and
> compile-test-error-regexps in that same
> test/automated/compile-tests.el).  To the extent that this code is
> likely to have parts that can be reused between different sets of tests
> for different packages, I tend to think that it falls in the camp of
> "it's just code like any other".
>
>> Anyway, I was just hoping that I wouldn't have to wade through years of
>> log history, as if porting test code wasn't boring enough already.
>
> Is the above rule "lax enough" to significantly lighten your workload?

No, I'm afraid not; it's really about the code that runs the test.

So I guess I'll just have to look through the history; I guess/hope that
Eric has written almost all of it anyway. The main problem is that we
weren't very careful with the tests when migrating to bzr, so I'll have
to checkout the old CVS repository (and the crowd goes "Eww!", but I'm
sure they just marvel at Lars' shiny new browser).

-David

Re: Tests and copyright

[Stefan Monnier]

> I guess/hope that Eric has written almost all of it anyway.

That sounds very likely.  Maybe he even remembers,


        Stefan

Re: Tests and copyright

[Eric M. Ludlam]

On 08/15/2013 12:48 PM, Stefan Monnier wrote:
>> I guess/hope that Eric has written almost all of it anyway.
>
> That sounds very likely.  Maybe he even remembers,

As far as I remember, most of the tests were written by me, or in some 
happy situations, but an author who needed to sign papers to get their 
changes into EIEIO in the first place.

I also put comments in the code or change log if someone contributed a 
test, such as from a bug report, of which I do see 1 example, and that 
example is not part of a harness, it's just a raw unit test.  Most 
examples in bug reports tended to come in a form unsuitable for the test 
suite.

I looked at the logs in the CEDET/BZR repository, and they appear to 
have translated from CVS ok to me.  Am I missing something?

Eric

Re: Tests and copyright

[David Engster]

Eric M. Ludlam writes:
> I looked at the logs in the CEDET/BZR repository, and they appear to
> have translated from CVS ok to me.  Am I missing something?

You are correct. I simply typed 'bzr log tests/eieio' and that ends with
the merge of 'newtrunk' because we re-structured the directories. The
files themselves have full history.

-David

Re: Tests and copyright

[David Engster]

Eric M. Ludlam writes:
> On 08/15/2013 12:48 PM, Stefan Monnier wrote:
>>> I guess/hope that Eric has written almost all of it anyway.
>>
>> That sounds very likely.  Maybe he even remembers,
>
> As far as I remember, most of the tests were written by me, or in some
> happy situations, but an author who needed to sign papers to get their
> changes into EIEIO in the first place.
>
> I also put comments in the code or change log if someone contributed a
> test, such as from a bug report, of which I do see 1 example, and that
> example is not part of a harness, it's just a raw unit test.

You probably mean that part in eieio-test-methodinvoke from a certain
"darkman"? I'll remove that part.

Then there's this bit

;;; Jan's methodinvoke order w/ multiple inheritance and :after methods.

where I guess this is Jan Moringen, so it should be OK?

Also, could you please give copyright to those files to the FSF in CEDET
upstream?

-David