Mats Bengtsson wrote:
Excellent! I just hope that everybody who contributes with a new
feature takes the time not only to make a regression test but
also an illustrative example (and of course, some text for the
main manual).
That would be nice, but I'll be happy as long as they add one snippet to
input/tolsr/ . Once a week, I'll do any necessary cleaning-up of such
examples and file them into the appropriate subdirs. Of course, this
makes Han-Wen unhappy... to be honest, I'm not certain how to resolve this.
Yes, I was slightly dishonest in my initial defense of tolsr/ -- it's
true that we need somewhere to store snippets that can't go in LSR yet,
but in my mind it's _also_ useful to have a convenient place for
developers to dump examples of new features. Of course, I certainly
won't complain if developers send in text for the main manual... but if
developers want to spend more time writing code, I personally won't
complain as long as they add snippets to input/tolsr/. Some
documentation is better than no documentation, after all.
(unless you or somebody
else plan to regularly review all new regression tests).
No, absolutely not.
I don't really understand the strategy here (I think I have
complained earlier too).
1) Whenever I'm making new changes, the most important thing (in my
mind) is that the new stuff is clearly better than (or at least equal)
to the previous thing. This "snippet subsection" is equal to the old
system, so I'm happy pushing forward on this.
2) The snippets are available for offline use. I know that fewer and
fewer people do significant work away from an internet connection, but
I'm one of them. :)
3) Quality control. Materials on the lilypond website are subject to
rigorous and thorough testing by our team of highly-trained... ahh, who
am I kidding? :) Still, Cameron and I will be making sure that the
examples aren't broken at least. Once the bulk of the initial setup is
done, more polishing will happen. Since LSR is entirely open to random
users (who can also edit the "approved" snippets), I don't think it's
suitable as part of our official documentation.
As long as we all
agree that this is a temporary solution, waiting either for LSR
to support multiple LilyPond versions and always stay up to date
with the latest LilyPond releases or conversely for LSR to be fully
integrated with the LilyPond releases, then it's fine.
I don't think that either of those will happen. I agree that it makes
life a bit harder for Trevor and the like... but I think we only have a
dozen serious users who track unstable.
Cheers,
- Graham
_______________________________________________
lilypond-devel mailing list
[email protected]
http://lists.gnu.org/mailman/listinfo/lilypond-devel
