alt Hacker News> But "at work" you generally have these decisions made for you.
The idea that most employers make terrible decisions now, and amazing decisions back in the day, is plainly false. The author vividly recollects working at a decent Java shop. Even there I strongly doubt everything was amazing as they describe, but it sounds decent indeed. But plenty businesses at the time used C++, for no good reason other than inertia, usually in Windows-only Visual C++ 6-specific dialects. Their "build server" ran overnight, you'd check in your code in the late afternoon and get your compile errors back in the morning. The "source control" worked with company-wide file locks, and you'd get phoned your ass back to the office if you forgot to check in a file before leaving. Meanwhile, half the web was written in epic spaghetti plates of Perl. PHP was a joy to deploy, as it is now, but it was also a pain to debug.
If you care deeply about this stuff, find an employer who cares too. They existed back then and they exist now.
Replies
This is like economic growth: First bad, then upwards trajectory, then now in free-fall
You are describing it: Things in programing were bad, then suddenly all in the upside UNTIL it start to coming down.
Is not a refute of the problem. Is to pick a moment were both were bad, and like all the discussions about tech, MASSIVELY ignore that MASSIVE internet with MASSIVE money with MASSIVE backing is worse than before.
Is like people complaining that pistols on the wild west kill as do nuclear weapons, ignoring the massive difference in size and blast damage
Amen.
We love the idea that we don't have any agency in this field and we're constantly being pushed by the mean baddies at the top.
Did people really only compile once a night in the days of visual studio 6? There were pentium 2s and 3s back then.
The "terrible decisions" of yore hold no comparison to today's "terrible decisions". It's not the same ballpark, it's not the same sport.
> Their "build server" ran overnight, you'd check in your code in the late afternoon and get your compile errors back in the morning.
This. Let's keep things in perspective when people complain about long Rust compile cycles. And even that's a whole lot better than filling in paper-based FORTRAN or COBOL coding forms to be punched into cards in the computing room and getting back line-printed program output (or a compiler error) the next week.