On 7 Dec 2007, at 02:20, Graham Percival wrote:
Hans Aberg wrote:
When I made the first report, I had no idea what was causing it.
All I knew was that replacing the quadruplet with a rest made it
problem disappear, so it looked to me as though that was the
problem. You, with your broader knowledge of LilyPond bugs,
I wish that people would stop assuming that I know what I'm doing.
For bugs, I simply spend five minutes experimenting. For example,
starting from this:
\relative c' {
\key d \minor
\time 12/16
\times 3/4 {g'8 f8} f4 \pitchedTrill f8. \startTrillSpan ges f8
\stopTrillSpan |
}
I did the following.
1) Is there a bug? ...compile, view... ok, there's no flat on the
ges.
2) The \times looks irrelevant; what happens if I replace it by g4
f4?
... compile, view... ok, bug still present.
3) What about the \time? That looks really silly. Let's remove
that, and change all the durations to 4.
... compile, view... ok, bug still present.
4) Key signature? remove, compile, view. Bug still present.
5) What about removing the \pitchedTrill ? remove, compile,
view... ah, of course. Bug not present, but neither is the ges, so
that was silly. Let's put the \pitchedTrill back in there.
At no point did my supposed knowledge of lilypond come into play.
All I ever do is experiment. I mean, I don't even know how
\pitchedTrill and \spanner stuff works, as was evident from my
trying to deleting them.
Oh, also by "remove", I mean "comment out, compile to test, and
only actually delete the item once I've verified that I still see a
bug when it's commented out". I learned this trick the hard way.
Sure you have a lot experience with structure of LilyPond bugs, you
even describe that it several places. If you want to cut down on your
own time, drop off hints as you did to me first time: "Is this or
that really necessary?" Then people not only can go on with reshaping
the example, but will learn how to do it for you. I have been on the
Bison lists for a decade, and it is common to deal with bug reports
where people send whole programs. To Apple, I have in some cases sent
in whole DVDs. One files a bug report, and the people set to handle
the bug should be able to indicate how the example should progress.
In some cases, I set up on their suggestions, I set up my computer
running for days. So if you want have bug reports have special form,
you need to find effieicnt ways to communicate that. In the case
above, I took out two notes of the quadruplet, and it did not change
the bug, so it was reasonable to assume it was the quadruplet causing
the bug, nd getting your opinion about that; only when I typed the
letter, I got the idea it might be the g. You could then use the
suggestion I made to further. This kind of interaction is typical for
bug reports. I have also done bug reports for compilers since the
early 1990s.
Hans Aberg
_______________________________________________
bug-lilypond mailing list
bug-lilypond@gnu.org
http://lists.gnu.org/mailman/listinfo/bug-lilypond
