Re: [PATCH] Document %define lr.type and lr.default_rules.

[Joel E. Denny]

On Wed, 6 May 2009, Akim Demaille wrote:

> Le 21 avr. 09 ? 12:32, Joel E. Denny a ?crit :
> 
> > +** IELR(1) and Canonical LR(1) Support
> 
> 
> I think that such one-person major contribution (like this one and the
> forthcoming changes from Alex) deserves to be credited personally in the NEWS
> file.

That's fine with me, but we should go back through recent NEWS entries and 
credit a few others: Paolo Bonzini, Wojciech Polak, Satya Popuri, and Bob 
Rossi.  Did I forget anyone?

While we're on this topic, I dislike that some files have the unstable 
"written by" comment.  Is there a legal reason for this?

From what I've seen, almost every major contribution to Bison has not 
remained a one-person effort, and I see that as an important testament to 
the significance of the contribution.  Thus, I think the "written by" 
comment belittles the improvements made by others, and I think it 
belittles the significance of the contribution itself.

I also find it arbitrary that credit is given per file.  First, code 
organization evolves.  Second, some equally important contributions don't 
come neatly packaged in distinct files.

If NEWS, THANKS, ChangeLog, git log, and the mailing lists are not 
sufficient to give credit to contributors, then I think we should at least 
change "written by" to "originally written by" or "originally contributed 
by".  I would also need to add my name to the IELR-specific files.

> I also wish we had some "History" section.  I enjoy very much what Alexandre
> Duret-Lutz did for Automake for instance, and what David MacKensie originally
> did for Autoconf.

That would be nice.

> > In the manual, the discussion of %define lr.* is getting too verbose for
> > Decl Summary.  I should move the details to a new section, but I haven't
> > figured out where yet.
> 
> I agree.  And more generally, we should spawn a whole section for variables,
> Var Summary, instead of staying in Decl Summary.

That sounds good.  Should we do this for %code as well?