Re: [g-a-devel]Commit permissions to at-spi ...

[Bill Haneman <bill haneman sun com>]

On Thu, 2002-08-15 at 10:59, Michael Meeks wrote:

> 	I don't want to have this argument; but for the record, a large ammount
> of my work was of this form of code removal eg.

...
> 
> 	Doing lots of cruft removal, and re-writing; 

Yes, absolutely true.  Your editorial work was extremely helpful and
welcome, as were your substantial other code contributions; no argument
there.

> I don't believe your
> argument that Marc and you are the only primary authors holds water
> under time-consuming deep scrutiny. But who cares.

We are the current maintainers/custodians, and the "original
architects".  I suppose lots of bits could be wasted discussing the
significance (or lack thereof) of that second attribution, in the
collaborative-software world of refactoring and iterative development.

> > I agree that you have made substantial contributions to the code base,
> > however.  Our policy to date, which is in line with what I was taught in
> > Copyright class back in the dark ages, is that mods to existing files
> > are treated as "derivative works" and thus continue to hold the original
> > copyright notice, whereas newly committed files carry the copyright of
> > the new author(s).
> 
> 	Common sense requires and common courtesy suggests that any non-trival
> change carry a joint copyright notice. Talking to lawyers is just a
> waste of time and money, bringing acute paranoia and expensive middle
> mangement problems. [ it's also excellent for the post-it note and memo
> industry ].

Examination of the source code shows that joint copyright is the
exception rather than the rule in GNOME at the moment.  I have no
fundamental objection to it, and I don't get my kicks talking to lawyers
either.  However I am not happy changing my own copyright-attributing
policy without at least one of the following:

	* there is clear consensus in GNOME/FSF as to how/when to add copyright
attributions;
	* my employers or their lawyers tell me to do it (and their
instructions seem compatible with FSF) ;-)

To date neither of these have happened.  The fact that the issue is
being raised is why I mention need for clarification.  Also, I think we
would agree that it's not appropriate for every person who commits a
bugfix or line of code to add a copyright attribution, so somewhere
along the line we need to be explicit about when to do that.
 
> >   We followed this convention, for instance, in the
> > Java Access Bridge for GNOME's file "unknown.c", which was originally
> > written by Erdi Gergo as part of "monkeybeans", though we ended up
> > replacing almost all of the code.
> 
> 	It should be a joint copyright statement; either way, there are several
> files in at-spi that are substantially my own work, added to the project
> by me etc. that bear only a (bogus by your argument) Sun copyright
> currently. I personally think that's an honest mistake, so I don't care
> as long as it's fixed.

Agreed.
 
> 	What I do care about, is that code that I wrote has stuff committed by
> you - Bill Haneman - without proper scrutiny, or due thought. That
> subsequently, I get called upon to figure out the wierd and wonderful
> problems in the code, and then I can't commit fixes to what is
> essentially 'my' code. [and it appears that I'm the only one who
> understands it - ironicaly, despite my efforts to write for clarity].

I think the less we use phrases like "your" code and "my" code, the
better off we will be.  

As for the "due thought" argument, I will say "mea culpa" in some
cases.  However I do not think I should be subjected to public flogging
as a result, we're all under pressure here and most of us are human ;-) 
We are *all* busy finding weird and wonderful problems on a daily basis,
trust me that you are not the only person having to root these things
out.  

Things will break, that is a given.  Reducing the frequency and duration
of breakage is a noble goal.  Personally I think that effective means of
testing is the weakest link in this particular chain, but that's a topic
for another email.

The subject of this email is "Commit permissions to at-spi".  As
everybody is aware, a (non-maintainer) commit was made recently that
wasn't actually reviewed; another honest mistake which we don't want to
repeat.  Though I see some merit in going to a more conservative commit
policy (i.e. all commits of substance require review), it would have the
effect of significantly slowing our development and bugfixing, and
probably delay our accessibility solutions.  Asking you to review my
patches in addition to my reviewing yours will obviously slow things
down.  

In the meantime I propose that we do this:

* all changes to at-spi are created as patches and posted to
gnome-accessibility-devel

* maintainers wait for some period of time (say, one day) for comments
before committing changes that functionally affect code

* all patches are subject to maintainer approval. 

The last point is current policy, the first two would codify something
we should have been doing anyhow, but which are a slightly more
conservative than most GNOME modules.  

> > I will seek some clarification on this; bear in mind that UK/EU and US
> > policies may not have harmonized fully on some of these points.  At any
> > rate you are given prominent mention in AUTHORS; however I should
> 
> 	Your mail runs out here ... but who cares about EU/UK vs. US law I
> suspect you'd do better doing some hacking / bug fixing. If Sun
> desparately want to hold copyright on at-spi, for reasons utterly opaque
> to me, I would have been happy to assign copyright before I started, I'm
> not a bigot in that sense, but now ...

No, no, not my point.  See above.  Who knows, we may want to assign
copyright to 'GNOME Foundation' or something someday, I don't care as
long as the AUTHORS file gives credit and the source freedom is
preserved.  I have no reason to believe that 'Sun' feels otherwise.

(BTW I would love to be hacking/fixing instead of so much email)

> 
> > s/contributions/major contributions
> 	Hmm ?

Substitute 'major contributions' for 'contributions' in AUTHORS.  A
trivial point, but more accurate.
 
> > > 	My patch (to follow) will fix some of the modules almost exclusively
> > > written by myself; But it would have been nice to have the greater
> > > contribution acknowledged.
> 
> 	It would be good to have my patch reviewed / committed, and would fix
> some of the most glaring travesties, along with making the code more
> efficient and functional.

Your flair for hyperbole is undiminished...

I may look for you on IRC if I have questions about the patch.

Regards,

-Bill
 
> 	Regards,
> 
> 		Michael.
> 
> -- 
>  mmeeks gnu org  <><, Pseudo Engineer, itinerant idiot
>