[_sh]

My thoughts on the philosophical aspects of the article don't really relate to Rust.

> These values reflect a deeper sense within me: that software can be permanent

Maybe it can be permanent, but it shouldn't be. Software is disposable and all the time developers are mistakenly fighting this aspect of its nature instead of embracing it. Great software is malleable and develops, over time, to adapt to the human that is using it. But it is impermanent--parts are snipped off here, fleshed out there, nothing stays the same, it is obsolete as soon as it is released. Don't delude yourself that your software will run for a thousand years, be like Warhol and celebrate the ephemera that is pretty much every program ever written.

> I have believed (and continue to believe) that we are living in a Golden Age of software, one that will produce artifacts that will endure for generations

Museum pieces, sure, but do we still want to be using generations-old software in the years to come? Hope not. Times change, needs change, and software that doesn't change is replaced by software that does, and quick. What about such monumental artifacts as 'cc', 'awk', or even the UNIX kernel? For years they have dominated the landscape, they are the Ozymandias, the King of Kings. If we are still clinging to these titans in another 20, 50 years, is that a good thing?

[ethelward]

> do we still want to be using generations-old software in the years to come?

*BSD, vim, Emacs, Perl, C, Apache & Linux in some way, gcc, the GNU userland, Air company & banks infrastructure come to mind, and they still do the work.

[_sh]

There are a couple of ways to look at this. Philosophically, to pick any one of your examples, my emacs is version 26.1. Is this the same software as emacs 15.10 released April 1985? Will a perl 6 program run on perl 5? Am I the same person I was 5 years ago?

Another take: what is the ratio of the same software still in use after, say, 10 years to all software in use? I would argue that more than 99% of all software (e.g. by version number) is no longer in use after a mere 5 years. Software is inherently disposable, let's not pretend we're building bridges that will stand for generations.

My point is that developers (I'm one) have a hard time with the qualities of software: we don't understand the nature of software change, and bicker about what bumping a semver number means, and we fight its disposable nature by engineering it to the point where it could run for a decade (it won't).

[jazzyjackson]

I'm glad there are Warhols doing their thing but I think the world would be worse off without Michelangelo.

awk, vim, whatever Dos 3.1 GUI the guys at B&H use to fulfill my orders, why shouldn't I expect to be using these programs in 50 years? They solve a problem and work every time.

[_sh]

> why shouldn't I expect to be using these programs in 50 years?

No reason at all, if they still solve the same problem, they're still the best solution. The problem with 'problems' though, is that once they're cornered they tend to change into a whole new problem. A nasty one the current solution won't work for, usually.

[scns]

COBOL?

[_sh]

SNOBOL!