Hacker News Reader

Comment by wiremine | Hacker News Reader

> exposure to these hidden problems

I think this is a key to experience. I spent a large part of my 20s diving into problems and totally screwing up the solution. I learned a ton from this.

On the flip side, maybe 50% of those problems don't exist anymore, because new technology or methodologies have solved them. Knowing how to use CVS or SVN doesn't really count for much in a world dominated by git.

I think that's where I value a deep understanding of CS and core engineering practices, regardless of age. Understanding pointers, how to debug, and the difference of the heap vs. the stack hasn't changed much over the years.

View on HN

> Knowing how to use CVS or SVN doesn't really count for much in a world dominated by git.

I don't understand the point you are tying to make. What has git so fundamentally changed that SVN/CVS don't count as knowledge / experience?

Additionally, who says git is dominating? Here, on this forum, it has popularity. In industry its almost unknown. And guess which is the larger part of the world?

> In industry its almost unknown.

Maybe I'm in a microcosm, but everybody in industry I work with switched to git years ago. And I'm definitely not in the valley. :-)

There's danger of selection bias. Would you work for a company that used a different tool?

> Would you work for a company that used a different tool?

Of course.

Most companies I've seen when freelancing as a web dev in Germany either used git or were in the process of migrating from SVN to git.

That said, there's selection bias involved.

A few data points to consider:

1) I happen to work in a dev tools company and we definitely see momentum towards git in our customer base. Broadly speaking 2/3rds of our customers (SMEs, F50s, etc. across a broad array of industries) are using it wholly or migrating to it.

2) Several of the larger proprietary SCMs are debuting git compatibility (e.g. p4), implying a belief that expending engineering resources along those lines is perceived as worthwhile.

3) The stackoverflow dev survey that came out recently had git with ~69% of the market: http://stackoverflow.com/research/developer-survey-2015#tech...

4) GitHub (just part of the git market) has over 10M repos now: https://github.com/blog/1724-10-million-repositories

You can always hope for more data, up to the limit of every programmer on earth answering your question(s), but the trend in the available data is clear: git is extremely popular and becoming ever moreso.

Every company I have worked for, since 2001, has used Perforce. My latest employer was starting to think about maybe possibly using something else, but I doubt it will happen.

I suspect this is a generational thing. The companies I have worked for were generally formed in the nineties or 00s, when the free source control systems didn't exist or weren't yet popular. The companies being formed now use Git almost universally.

I've been writing since the late 80's professionally and had to look up Perforce. Mostly we used stuff we created until the late 90s and then it was Subversion. Once dragged to git about 10 years ago I would never look back. We do host it ourselves for security using atlassian stash though. Love it.

Because Perforce is fantastic. Nothing else comes close. The chance to work at a Perforce shop again might even lure me away...

Those stupid old guys eating steak, when all the cool kids are eating burgers...

In industry its almost unknown.

So unknown that even Microsoft offers git functionality in Visual Studio now: https://msdn.microsoft.com/en-us/Library/vs/alm/Code/git/ove....

Maybe it's not dominating (or maybe it is, don't care), but to say that git is almost unknown is a bit strong, IMO.

> What has git so fundamentally changed that SVN/CVS don't count as knowledge / experience?

I think git (and other distributed source control systems) changed the culture around branching and forking. So, if you're coming from a CVS world, there is a lot of technology that changed, and the technology changed the culture/methodology.

I.e., if you dropped a CVS user from 1996 into today's world, they would have culture shock, not just a new technology to learn.

All source control tools have a different model of the source history. You have to relearn that to change tools.

The culture change mentioned is nothing to do with Git. It has to do with GitHub. We can talk about that too if we want to.

Of course, git was developed by an old guy that used CVS, so....

Replace CVS with BitKeeper, and you'd be right.

Linus never used CVS because he couldn't stand it. The kernel had no version control until BitKeeper came along specifically because Linus wasn't sold on version control until he discovered BitKeeper, and then when McVoy killed their licenses, Linus chose to write his own system because he didn't like any of the existing ones.

I must be getting old

But BitKeeper was created by an old guy (Larry Larry McVoy[1]) who was born in 1962. Not only that, but BitKeeper is built on top of SCCS[2] which predates RCS (and RCS predates CVS).

BitKeeper was a totally new solution to a very old problem that was "solved" years ago. It was created by an "old guy" who knew the wheel had a large flat spot and figured out how to make it round. It was subsequently adopted by another "old guy" (Linus) who refused to use wheels with flat spots.

[1] https://en.wikipedia.org/wiki/Larry_McVoy

[2] https://en.wikipedia.org/wiki/Source_Code_Control_System

Are we confusing age and culture with stupidity?

A stupid person wouldn't be able to adapt, or might have trouble adapting from one tool to another. That's a stupid person. Likewise a stupid person might think that an organization that was shipping successful products and making money was making a big mistake by not migrating off of svn.

Stupid is stupid, age is a different dimension. "culture" especially regarding tooling like this is sort of a cop-out.

It's true! I did a fair bit of programming and other computery things in undergrad and had a pretty good track record for, say, transforming some trash from behind the CS department into a workable computer. I was happy with Emacs, resisted vi, could whip up a nice html page with some cutting-edge CSS 3. By no means did I have special talents but I was sort of up to date on a broad range of basic computer-related skills, from hardware to networks to web design.

But then... I went to grad school in pure math. Seven years later, I emerged from a rip van winkle sleep to a whole new world. Still hand-writing my html, like all mathematicians do, I was astounded to wake up to a world with... frameworks?! AJAX! Rails! widespread use of IDEs, an IDE for every language or taste, Javascript everywhere... Python and Ruby as non-fringe languages....

It was a bit of culture shock and it was fun.

The major concepts are still the same.

Yeah the hidden problems I am talking about are less specific.

Some of them are problems like workers unions standing in the way of automation of some shipping procedure or that people never thought about it as a problem but rather the cost of doing business.

Technology definitely removes a lot of those barriers over time as you say. But there are still plenty of problems that aren't technological and perhaps those are the most interesting.

> But there are still plenty of problems that aren't technological and perhaps those are the most interesting.

100% agree. A lot of problems involve navigating a particular business domain or learning how to work with people.

Reply on news.ycombinator.com