Florent Becker writes: > Le 09/11/2010 20:20, Stephen J. Turnbull a écrit :
> Yes, I should not have pointed at the contributors, but the market-share > is a real difference. To say it otherwise, Python or Bazaar have (or are > closer to) a critical mass of users that ensures the current developpers > stay motivated (or get paid to contribute) In my experience, numbers of users don't motivate developers very much. Sometimes their feature requests are interesting, and bug reports hit at our pride. But *nobody* (and especially not devs who claim that their features are demanded by users :-) goes out and does the marketing research necessary to find out what users really need. Pay is another matter. But guess what? Users don't pay for software development. Angels do, and (for better or worse, but definitely more fun :-) most software angels are driven by developers or ex-developers, not users. > and that they have prospective new developpers (we're not doing too > badly on that count, but I think it's a part where innovating is > important). That may be. But I think that (aside from the practical importance of "we do cherry-pickin' right") the use of a modern (even if it's rapidly becoming The Establishment) functional language and semi- rigorous correctness proofs are sufficiently innovative (even five years later) to attract a lot of developer interest.
> That critical mass of users also means that they get more indirect > benefits from third parties, like support in IDEs~ If eclipse does > not support git (does it?), it's eclipse's problem. It does, and it's not. Regarding benefits from third parties, Eclipse supports a plugin architecture, and third parties (users) wrote the git support. Not the git people and not the Eclipse core. > If it does not support darcs, it's our problem. Realistically, yes, lack of Eclipse support would be a blocker for many potential Darcs adopters. But there are plenty of other blockers for Darcs, unfortunately, and there are a couple of prominent examples of projects that got sick from eating the dogfood. Now, *that* hurts. (I know that for a fact because Darcs was mentioned in the dVCS transitions of XEmacs, Emacs, and Python, and got a cold reception -- everybody who mattered was aware of the Haskell repo issues.) I don't think either issue is particularly a blocker for the likely contributors, though: they're more interested in hacking Haskell or patch theory, as long as it works most of the time. > > On the one side, the statement is false; there are plenty of not > > terribly innovative projects that do quite well: take the Emacsen for > > one example. > > Yes, but they have a critical mass of users. By and large ignored by the developers. It's not a motivator, and the users as a group don't contribute much. It's developer usage that matters. > > On the other side, it's trivial to show that Darcs is > > where some VCS innovation happens: Darcs is still the only VCS that > > can say "we do cherry-pickin' right!" OK, David did that and you > > can't really claim credit for that in the 2.x series, but that's a > > developer issue. > > It's a capital we cannot live on eternally. No, but it has a half-life of at least 5 years. > > Ask your users, see what they say. If Zooko says "Darcs isn't > > innovating fast enough, and I'm thinking about switching to git for > > that reason", I'd be worried, too. But I don't think he'll say that. > > Well, he might very well say "oh, I want eclipse integration", Well, I don't recall him saying that. I know that Eclipse integration is a big thing in the Java community because of the "file name must be derived from class name" thing that really imposes a lot of scutwork on would-be refactorers, and because of the really crappy style :-) that Java seems to impose on its users. But Zooko's concerns as seen here are mostly stability and performance, as I recall it. Not features. I could easily be wrong, and I'm slightly out of line for trying to guess what Zooko would say, I admit (but I want to close this off; I don't really have anything more to say at this point).
> and kiss us goodbye once a coworker shows him hg has got cherry picking a bit > less wrong than it initially did¹. Only fanboys value innovation per se, > but if new features do not come to darcs, and come to other vcs, and the > cost of switching goes down, then at some point, the cost of > not-switching prevails. > > Also, as I said earlier, showing that new stuff happens is vital in > terms of getting new developpers. Why don't you poll the developers you have and find out? I'll bet that the honest ones will tell you that it doesn't matter how much innovation others do, it's how fast their own new stuff gets into the mainline that matters to them. ;-)
_______________________________________________ darcs-users mailing list [email protected] http://lists.osuosl.org/mailman/listinfo/darcs-users
