31–40 of 255 posts
It's also very simple to create packages, it took me less 30 min to learn how to create a package and to deploy it. In comparison last time I checked, creating deb packages is absurdly complex, and I usually just give up and provide a Bash file instead.
Earlier quoted context omitted.
It's hard for me to understand the issue with this. Google didn't hire the guy who wrote software that had a big impact on the community. Does it mean Google misses good engineers because of their less-than-ideal interview process, or simply Google hires only people who are a good fit for their organisation? Writing the overly bitter tweet surely won't help with improving hiring points. Seems highly unprofessional an…
Well, he's at Apple now in charge of the Swift Package Manager, so I think he did okay.
Earlier quoted context omitted.
It's hard for me to understand the issue with this. Google didn't hire the guy who wrote software that had a big impact on the community. Does it mean Google misses good engineers because of their less-than-ideal interview process, or simply Google hires only people who are a good fit for their organisation? Writing the overly bitter tweet surely won't help with improving hiring points. Seems highly unprofessional an…
Well, he's at Apple now in charge of the Swift Package Manager, so I think he did okay.
As a package maintainer, though, one thing surprised me: you can't make your package depend on a specific version -- even a major version -- of a library.
So if MyApp uses YourLib 1.0, everything is fine until YourLib 2.0 comes out, at which point users doing `brew install myapp` will start getting cryptic compiler errors. I have no choice but to drop whatever I'm doing and make MyApp compatible with YourLib 2.0, which may even be impossible.
It's just odd because otherwise Homebrew is so well designed. It seems like a weird omission.
Edit: Some packages get around this by packaging major versions separately (qt, qt5) but most don't.
Among all the software I use, Homebrew must be the one giving me the least trouble. I have 100 installed packages and I don't remember the last time anything didn't work. Plus, for anything I care about, updates seem to always take less than a day, usually it's almost instantaneous. In comparison with MacPorts before, it's like night and day. npm/pip/mix/bundler all seem to periodically conspire to waste a few hours…
Some design choices still rub me the wrong way to this day though (quibbles to some, but for a couple use cases it can matter), that's why I'm trying hard to bring back Arch OS X to life.
Anyway, congrats to the Homebrew team! They truly deserve it.
Throwback to that one time Google wouldn't hire @mxcl. ;) https://twitter.com/mxcl/status/608682016205344768 Well done Mike and contributors! It's a great piece of work.
He couldn't even ${college_level_gotcha}! No wonder he wasn't hired!
For instance what if they code in such a way that they haven't run into that problem in years as it is internalised. Or maybe they make use of tools that eliminate a lot of problems, such as TypeScript or similar.
A good developer comes in many forms and increasingly often from all over the world.
I'm excited about Homebrew potentially coming to Linux. When I first started using OS X more regularly, I was surprised by how better an experience OS X + Homebrew is compared to Ubuntu/Debian repositories + PPAs/third-party repositories.
As a user, Homebrew is a great experience. As a package maintainer, though, one thing surprised me: you can't make your package depend on a specific version -- even a major version -- of a library. So if MyApp uses YourLib 1.0, everything is fine until YourLib 2.0 comes out, at which point users doing `brew install myapp` will start getting cryptic compiler errors. I have no choice but to drop whatever I'm doing and…