> Do y’all not long for a future where you can get off the upgrade treadmill because the developer facing API is fixed? Not backwards compatible because that implies you ought to be moving to the next, like once it works you can call it done.
No. We want modern APIs that prioritize modern concerns, usable with modern toolkits and frameworks and taking advantage of modern programming principles. Even if we had an API intended for longevity, once its creators die we will tear it down completely and replace it with something that suits our newer-therefore-better tastes. For them it was the cornerstone of an industry's worth of innovations; for us it is but a millstone around our necks and must be replaced. I know this because I've seen it happen many times. Once the people who've staked their entire careers building upon $THING, and developed some truly remarkable software, grow old, up rises the chorus of people who are sick of $THING, who can't even fathom how anyone got anything done with $THING, who give talks at conferences about how $THING is fundamentally broken and how we should be using $NEWTHING instead. And these voices grow louder, their chants more thunderous, until it's generally accepted that $THING is a relic and $NEWTHING is the future. Even the things we thought would last forever -- POSIX, C, X11 -- are now, if anything, well past their expiration date.
This is how things are. This is how they must be. There is naught we can do but be like a Japanese person observing the seasons, contemplating, with some sadness, the constant change and the endless cycle of death and rebirth.
There's such a thing as lasting too long without deprecating old cruft.
POSIX isn't too bad, but is still kind of a mess. C has a lot of broken features. X11 doesn't fit how hardware works these days and trying to force it with extensions isn't a good fix.
There are real problems with faddishness on the scale of 5-10 years, but there are also real improvements that happen across decades, and if 40 year old tools don't get updated then they should be replaced.
But you forgot to mention how this requires ignoring the old devices. Can X11 replacement not work on old hardware? What critical improvement in POSIX would mandate the same?
Unless that's slowly=unusable that's not a good example (would also help if this poor support was a result of some "modern programming principles"). Otherwise that's way too early to retire in Japan
No. We want modern APIs that prioritize modern concerns, usable with modern toolkits and frameworks and taking advantage of modern programming principles. Even if we had an API intended for longevity, once its creators die we will tear it down completely and replace it with something that suits our newer-therefore-better tastes. For them it was the cornerstone of an industry's worth of innovations; for us it is but a millstone around our necks and must be replaced. I know this because I've seen it happen many times. Once the people who've staked their entire careers building upon $THING, and developed some truly remarkable software, grow old, up rises the chorus of people who are sick of $THING, who can't even fathom how anyone got anything done with $THING, who give talks at conferences about how $THING is fundamentally broken and how we should be using $NEWTHING instead. And these voices grow louder, their chants more thunderous, until it's generally accepted that $THING is a relic and $NEWTHING is the future. Even the things we thought would last forever -- POSIX, C, X11 -- are now, if anything, well past their expiration date.
This is how things are. This is how they must be. There is naught we can do but be like a Japanese person observing the seasons, contemplating, with some sadness, the constant change and the endless cycle of death and rebirth.