I'm really just a fairly uninterested bystander who's been eavesdropping. I don't much care one way or the other, except that I'd *love* to see Pharo explode into the popularity that it deserves.
On Sun, Nov 3, 2013 at 9:34 AM, kilon alios <kilon.al...@gmail.com> wrote: > Influence the way Github goes is besides the point at this point since we > dont have a better alternative and I will be a great oracle to predict , or > not, that we wont have one in the next decade. Alan Kay once said something along the lines of "The best way to predict the future is to create it." > Unless something > extraordinary happens and Pharo goes super popular. I think that may be the wrong attitude. Even if it doesn't, the best ideas will filter down to inferior languages/environments. Things like Visual Studio and Resharper are still trying to bring Smalltalk into the mainstream...though in a very limited scope. If someone creates something that's a significant improvement to github, and it's generally useful, it won't matter what technology created it. It either will supplant github or github will extend and embrace the ideas. Either way, the ecosystem improves. I could probably write a thesis about the actual economics and morality and sanity of using closed source created by a company like github vs. something like squeaksource (which I loved), but (AFAICT) the community is faced with exactly that sort of crisis. It takes time and effort to maintain and improve anything. People who are making money doing so are in a better position to make that "anything" nicer and friendlier to end-users than people who are working on it in their/our spare time. It's a sad fact of life: most of us don't get to be Richard Stallman (or maybe that's a happy fact...whatever). > Being a commercial application has not stopped tons of open source projects > porting to it. But hey what do they know. I am not sure what you mean about > sourceforge, I see still loads of open source project using it. Have you checked out sourceforge lately? It was obnoxious for years. Which is probably the main reason that so many projects have moved to github. The last time I tried to download anything from it, I had to wade through 3 pages of "Download and install this malware to do things you already can!" searching for the "Never mind, just let me download the thing I want" link that was buried in all the crap. In the end, I *still* wound up with 3 things I didn't want. I can only assume they'd have tried to auto-install if I were running Windows. I found out the next day that they'd been acquired by some new company. I intend to avoid them like the plague from now on. Github might wind up in the same boat, but I think that would require some major redesign on their part. It seems like it'd be pretty tough to insert adware into the `git clone` process. > And if Git is good enough to version control Linux source, I dont see why it > would be a big problem to Pharo. Several years back, when people were starting to realize that tools like Source Safe and SVN suck, I was involved in a long conversation sort-of along these same lines in one of the lisp communities. Everyone was looking for the next big thing in source control. At the time, Darcs looked like the thing that should win, but wouldn't. Git was too big, nasty, scary, and didn't work on Windows, so it didn't even seem to be a contender. Every option available was focused around old-school line-based diff's. Which seemed totally asinine from a lisp perspective: that was really about modifying leaves and branches in a tree. I probably got flamboyant about this at the time. Finally, someone politely suggested that I go ahead and write it myself if I thought it was such a great idea. So I tried, and it turned out to be a lot tougher than I'd expected. Most of what I learned from that experience was the basic lesson that I didn't have a clue about lisp. But the experience seems at least a little relevant here. Git seems like a horrible tool for tracking changes to a smalltalk...I dunno. I'm such an outsider and newb here that I'm not even sure about the proper terminology. Package, maybe? Whatever. Group of classes that you write and maintain that implements some sort of useful batch of functionality. At the same time...would it be worth the effort to implement something in the same realm that would be competing with an established opponent? Then again, sooner or later, *something* will come along that makes git look as obsolete as git did to CVS/SVN. > But even if it is should not be faster if > we make extensions to git via monticello to work around these potential > problems instead of doing a huge project from scratch ? FWIW, I'd vote for that sort of approach. If I had any right to a vote here. Evolution beats revolution. Then again, I've dabbled around with what that actually means (I've been trying to come up with a decent way to visualize branches in my very limited spare time), and git seems to be pretty nasty to deal with under the covers. > I think being against git is perfectly fine, wanting to trying your own > thing is awesome, focusing on pharo specific solution is of course a good > thing. But then not wanting to do that and use what is already out there can > be as wise. Git sort of arrived in the middle of a perfect storm. Everyone was feeling the pain of source control systems that truly sucked. Linus felt that pain and threw a solution together. Its victory may be another great example of the way "less beats more", or it might be a testimony to name brand inertia. Or maybe it's just that intellectuals don't understand the wisdom of the market. Be that as it may. I still get riled up enough about source control to write long rants like this one when I have better ways to spend my time. Git sucks, but so does everything else. I don't have the time to invest in making it better. I'm too busy working on problems that I consider more interesting. I suspect this is the simple reality of what's going on here. If someone *does* care enough about the ecosystem to truly invest in a VCS that makes more sense than git, well, power to you. In the meantime, it seems like building beautiful and intuitive UI's on top of github's existing architecture would be more productive. Hmm. Now there's an idea to explore. Going back into lurk mode. Respectfully, James
