I do feel good about transitioning from thinking of myself as a software engineer to thinking of myself as a software artist. But I'm not sure which one sounds like the more horrible pretentious ego claim, when I really feel about these things more that they're statements of intent, priority, mindset and aspiration.
Timeline
Post
Remote status
Context
1Like for example, watching Greg make drinks where he challenges himself to highlight the existing ingredients in a drink, use the best version of them that he can, make them look good without hiding any of them in embarrassment, and make an incredible final synthesis. It's like Chopped. This is a lot of what I try to do, in a different medium. That often results in good engineering, but it would be weird to call the process or motivation "engineering."
Replies
4Git, to me, is an example of a near miss, where it's Actually Good to highlight the fundamental data model (Merkle trees of files, commits tying them to metadata and history themselves being signed immutable artifacts, branches being mutable pointers into the CAS) but a lot of the porcelain on top is cursed.
The challenge is to be proud of the pieces, not add so much on top that they get lost, while still increasing the operator's leverage in a way that is itself humble and brief and explicable.
A lot of abstraction (and information hiding rhetoric) is writing off the idea that this is possible, and wallowing in shame about your ingredients.
Wait what's cursed about git? The only cursed thing I know is mergetool, and can you figure out a better way to fix merge conflicts? I sure can't!
Git does have fundamental data model problems, for instance packing (obviously) restructures the files, so they take extra long to fetch after being repacked. And a data model that could handle partial fetches without a lot of crazy stuff under the table would be nice. But all the visible stuff you use is just... clone, pull, add, commit. Maybe push.
Git does have fundamental data model problems, for instance packing (obviously) restructures the files, so they take extra long to fetch after being repacked. And a data model that could handle partial fetches without a lot of crazy stuff under the table would be nice. But all the visible stuff you use is just... clone, pull, add, commit. Maybe push.
@cy the ui of git 1.x was fucking terrible but it's a lot better now, I used to fucking hate it compared to Mercurial but now I like it fine. To be fair you really need to compare it to what came before it, current git is actually enjoyable by that standard. I never used bitkeeper though
Ah, I don't recall git back in the day. I might have been a late adopter.