@djaouen, the article starts off great and I’d love to read more, but as others have noticed, it’s probably counterproductive to share an article that’s behind a loginwall, especially when it’s your own writing.
I look forward to reading the full thing if/when it becomes available.
For me, I am astounded how VCS took over the world, which is something almost nobody used 25 years ago. I went through CVS->svn->git transition (and even a bit of hg in the meantime). Same for automated testing and running and editing code on a shared remote server vs running it locally (although that last thing seems to have come full circle with AWS and all that crap). I’m also very happy how Unix/Linux and open source went from fringe which only weirdos used to more widespread use.
To me it seems crazy that VCS was so primitive for so long. I remember that I didn’t even realize that the idea existed until ~10 years of basic programming. On a senior year college graphics project, my team-mate showed me how to use “CVS”. I thought it was fine, but not mind blowing.
At my first job, we used Perforce, and I loved it. I thought it was such a useful tool, and it had both a nice command line and nice GUI. My team’s codebase was orders of magnitude larger than anything I’d seen, and I hadn’t worked closely on a team like that, so I guess I just liked how you could see what everyone else was doing. It was also fast.
After Perforce, I used svn, mercurial, and now git.
I understand why many people are unhappy with git, but personally I find I just get a ton of work done with it. Maybe every 2 months a branch gets in a weird state, and I just create a new one, and that’s about it. I’m able to take PRs on my project pretty smoothly. I like that I can see what contributors are doing and patch up commits on a separate branch before merging.
Automated testing is also huge … we didn’t have it at my first job, and I can’t write code like that anymore. Although that attitude probably prevents me from working on hardware/software projects, because the hardware often introduces some testing headaches / slowdowns.
Ugh, this article is hidden behind a login prompt
It goes away if you close it. I’m more annoyed by the presumed-consent, not GDPR-compliant, cookie banner.
The prompt was unclosable before, the author seems to have changed something so we don’t get that anymore.
@djaouen, the article starts off great and I’d love to read more, but as others have noticed, it’s probably counterproductive to share an article that’s behind a loginwall, especially when it’s your own writing. I look forward to reading the full thing if/when it becomes available.
I have gone ahead and removed the paywall from the article. Enjoy!
For me, I am astounded how VCS took over the world, which is something almost nobody used 25 years ago. I went through CVS->svn->git transition (and even a bit of hg in the meantime). Same for automated testing and running and editing code on a shared remote server vs running it locally (although that last thing seems to have come full circle with AWS and all that crap). I’m also very happy how Unix/Linux and open source went from fringe which only weirdos used to more widespread use.
To me it seems crazy that VCS was so primitive for so long. I remember that I didn’t even realize that the idea existed until ~10 years of basic programming. On a senior year college graphics project, my team-mate showed me how to use “CVS”. I thought it was fine, but not mind blowing.
At my first job, we used Perforce, and I loved it. I thought it was such a useful tool, and it had both a nice command line and nice GUI. My team’s codebase was orders of magnitude larger than anything I’d seen, and I hadn’t worked closely on a team like that, so I guess I just liked how you could see what everyone else was doing. It was also fast.
After Perforce, I used svn, mercurial, and now git.
I understand why many people are unhappy with git, but personally I find I just get a ton of work done with it. Maybe every 2 months a branch gets in a weird state, and I just create a new one, and that’s about it. I’m able to take PRs on my project pretty smoothly. I like that I can see what contributors are doing and patch up commits on a separate branch before merging.
Automated testing is also huge … we didn’t have it at my first job, and I can’t write code like that anymore. Although that attitude probably prevents me from working on hardware/software projects, because the hardware often introduces some testing headaches / slowdowns.
git definitely has some problems: large files, large repos, can’t lock a file, non-associative merge, etc.
i still pick git every time, though. maybe i’d go for perforce if i was doing large-team gamedev, or pijul if it matures more
What do you mean by non-associative merge? Like, when is it a problem?
in practice, almost never
here’s an example, though: https://tahoe-lafs.org/~zooko/badmerge/simple.html