On Thu, Feb 28, 2013 at 6:13 PM, Joseph Rushton Wakeling <joseph.wakel...@webdrake.net> wrote: > On 02/28/2013 06:20 PM, Adam Spiers wrote: >> >> I strongly disagree, unless your definition of "difficult" ignores >> the time dimension of such a project. >> >> http://www.joelonsoftware.com/articles/fog0000000069.html > > It can go horribly wrong, yes, but it doesn't have to. Git for example was > a from-scratch attempt at DVCS -- if Linus Torvalds had started from e.g. > Arch or Darcs, it's unlikely that we'd have had the innovative DVCS that we > see today.
That's an excellent counter-example, thanks! FWIW, git is very roughly half the LoC of LilyPond. >> Know-how and experience do not enable large, functional code-bases to >> be magically constructed in short time spans. They help increase >> velocity and quality, but any new code-base takes a long time to grow. > > Yes, but it takes even longer if you begin from the wrong starting point. > Neither Lilypond nor MuseScore really does what this team seem to be aiming > for Well, presumably that's speculation unless you have inside info ;-) Or did I miss some publically announced details? >>> but because they are writing a completely new codebase, they do >>> not have to be constrained by historical mistakes or backwards >>> compatibility. >> >> Nor would they be constrained by these if they started with LilyPond. > > Lilypond's existing architecture is a constraint, its existing syntax is > another constraint. And much as I admire Lilypond, I doubt its design is > entirely free of mistakes ... ;-) Sure; my point was that existing architectures can be changed without a rewrite. >> Finally we can agree on something ;-) But Daniel Spreadbury already >> admitted that they haven't even looked at the LilyPond and MuseScore >> code, therefore they have dismissed the possibility even before doing >> a technical feasibility study. > > Well, of course not. They won't want to risk the possibility of GPL'd code > influencing what they write. The simplest legal defence against accusations > of copying is, "I've never looked at that code." Right, he already said that in his response to me. > But the bottom line is, their licensing choices and their reasons for > building a project from scratch are fairly orthogonal. Maybe; we can only guess what their reasons are. > Personally, I doubt that Lilypond's design choices fit well with a piece of > software that's designed to provide real-time WYSIWYG engraving, Of course there's not a natural fit yet - the most obvious clash being the number of engravings per process invocation. But things like that are surmountable through refactoring, and I believe people such as Jan and the Scorio guys have already worked on solutions to this. > MuseScore's data structures are likely to be far too limited compared to > what the former Sibelius team are setting out to do. Probably, but again, data structures can be changed ... > It's a shame that the opportunity to do so was taken away from them by a > firm corporate decision to build a proprietary project, We don't know that's what happened, do we? > but I suspect that even with firm backing for a free software > package, they'd have chosen to start from the ground up. Sadly I think you're probably right. I think I'll retire from this thread now, since there's no use crying over spilt milk for longer than necessary ;-) _______________________________________________ lilypond-user mailing list lilypond-user@gnu.org https://lists.gnu.org/mailman/listinfo/lilypond-user
