I see we've got yet another coder who doesn't know the value of time or money.
Assume each client project has three coders (you plus two others, perhaps at the client site). A small github account is $12/month. A small Lighthouse account is $10/month. That's $22 per month. At, say, $60 per hour -- not a very high rate -- that's 22 person-minutes per month.
Perhaps you can easily find a trustworthy, experienced system administrator who, for a total cost of less than $22 per month, will set up Trac and git servers, keep them up, back them up, keep up with the security bulletins, install security updates, migrate to new versions, take tech support phone calls and -- most importantly -- be independent of you, so that after the project is done you can hand the keys to the client and walk away without further commitment. I fear that I would spend more than $22 worth of my time just trying to compose the want ad for such a person. I might waste more than 20 minutes per month just worrying about my hand-rolled system.
one of the entire points of a DVCS is that you do NOT need a central repository...
Just because the technology lets you, me, and the production server push and pull commits directly from our coworkers' laptops doesn't mean that such behavior is sensible, secure, easy to manage, or easy to teach.
The teaching feature, in particular, is key. Github isn't just a git server... it's got features that make git easier to understand. At this stage, that's kind of important, since most of the coders I know don't understand git yet.
I agree with you that the author is overly negative. If github provides some value to some people, more power to them.
At the same time, I object to "doesn't know the value of time."
As an independent coder, I can turn two hours into a working Trac + git, or I can turn $22/month into a github + lighthouse setup, but depending on the context, it could be very difficult to turn an arbitrary two hours into $22. It's not always possible to just bill another two hours. That asymmetry is why it's sometimes better to DIY than to pay cash.
I'm glad that you understand that the problem with the author's post is not that he doesn't find value in github or Lighthouse. It's that he can't imagine how anyone could find value in github or Lighthouse.
And I don't want to argue that it's never better to DIY. DIY is an important learning tool: I've installed gitosis once or twice, and I've installed Trac once or twice, and that's good. And DIY is great if it's your hobby: I do a lot of my own cooking, for example.
But I'm afraid that, if you already know enough about computers to install git and Trac inside of two hours... unless you're marooned on a desert island with nothing but the git source, a statement like "it could be very difficult to turn an arbitrary two hours into $22" really is an underestimation of the value of your time. Here's some things you could do with two hours that are probably worth $22:
-- Bill half an hour at $45 per hour and then spend 1.5 hours playing WoW. Assuming that's not possible...
-- Spend an hour, spread out over a day or two, surfing the web and bidding on tiny one-hour projects (e.g. "install Wordpress for me") that pay $22. Accept one and earn $22.
-- Upgrade your client base: Spend two hours pitching new clients via email. Bid $11 per hour more than your current rates. Assuming you can make 1 bid per hour, and the average client project lasts 40 hours, your success rate need only be 5% to make $22 an hour doing this.
-- If you know Rails read some sites about Django. If you know Django read some sites about Rails. If you know both read a book on Drupal. Now, when someone in your network of friends asks "should I use Rails or Drupal for this project?" you can answer intelligently. The additional job leads that you'll pick up will be worth at least $22.
-- Write a tutorial on installing git and Trac. Put it up on your blog. Crosslink your blog with your resume. This will reflect positively on your value as a programmer and raise your future salary by at least, say, $1 a month. Twenty-two months into your next job, you'll have earned $22.
-- Flirt on Facebook (a task that is much harder to reliably outsource than git administration). Or try speed dating. Or take one dance class. By itself, this won't pay off. So follow it up with more of the same. Keep going until you meet a wonderful partner and move in with them, at which point you will realize that, while it took you several hundred hours to find that partner, by living together you're saving hundreds of dollars a year in expenses like rent, and maybe even taxes. Assuming you stay together long enough, your original two hours will end up being worth far more than $22... even if you're an utterly unromantic pragmatist. :)
At a certain point dvcs has to be somewhat centralized. If I'm managing an app in git and need to push to production, it's much easier to pull from a service than to try to setup something that allows production to pull from my development machine when it comes time to deploy.
doesn't sound stupid to me...@ github you are not just paying for git... you can do that yourself... you are paying for backups if your laptop / server gets hosed
Assume each client project has three coders (you plus two others, perhaps at the client site). A small github account is $12/month. A small Lighthouse account is $10/month. That's $22 per month. At, say, $60 per hour -- not a very high rate -- that's 22 person-minutes per month.
Perhaps you can easily find a trustworthy, experienced system administrator who, for a total cost of less than $22 per month, will set up Trac and git servers, keep them up, back them up, keep up with the security bulletins, install security updates, migrate to new versions, take tech support phone calls and -- most importantly -- be independent of you, so that after the project is done you can hand the keys to the client and walk away without further commitment. I fear that I would spend more than $22 worth of my time just trying to compose the want ad for such a person. I might waste more than 20 minutes per month just worrying about my hand-rolled system.
one of the entire points of a DVCS is that you do NOT need a central repository...
Just because the technology lets you, me, and the production server push and pull commits directly from our coworkers' laptops doesn't mean that such behavior is sensible, secure, easy to manage, or easy to teach.
The teaching feature, in particular, is key. Github isn't just a git server... it's got features that make git easier to understand. At this stage, that's kind of important, since most of the coders I know don't understand git yet.