3 ms·
A language must change, or it will be left behind, and become an esoteric toy. Even C has to deal with this. The only reason C is still popular is because of
by kbenson 10y ago
A language must change, or it will be left behind, and become an esoteric toy. Even C has to deal with this. The only reason C is still popular is because of it's grandfathered status in so much of our infrastructure, and in many of the popular OS choices.
A language that changes will alienate people, and lose people. Perl saw this. Python is seeing this. The way to guard against this is to keep backwards compatibility. That generally doesn't allow enough change in languages that provide enough constraints to make them popular for large engineering projects, whether open source or commercial. Lisp doesn't need to change much, it's malleable enough in many respects that you can implement whatever you need, but good luck getting large engineering projects done and maintained using it. Perl has the same problem.
The only solution I see to stay relevant is to allow alienating users, but make sure those changes that alienate users are good enough to draw substitutes, enough to keep steady state, or preferably still grow slowly. Maybe prior users will even swing back in occasionally, and without the false sense of betrayal, they'll actually find they like the state of the language a few years later. For a while, at least.
This seems like the end of the world to Python community members, and it seemed like the end of the world to Perl community members because of the prior status those languages held, where they are or were at the top of their respective niches. I doubt the Scala community worries about exactly the same things.
Put another way, did you expect to still be writing Python the same way in 20 years, with only the popular libraries changed? If you did, did you think about what that would mean for the language, and what the community would look like at that time? I'll tell you. Perl. That's not necessarily a bad thing. I write Perl professionally, every day. I love it. It's not the worst thing, being stable and reliable. But you don't get to be top language in your niche and sit on your laurels at the same time.
Change is good. Embrace the change. If you're lucky, you'll use many languages to program in your life and be happy. The only way I see anyone being happy using any single language forever is to bury their head in the sand and ignore everything else going on around them.
- joshlemer 10y ago>I doubt the Scala community worries about exactly the same things. Why do you single out Scala as being particularly different to Python and Perl?
- chongli 10y agoEven C has to deal with this. But is there any C11 compiler out there that will refuse to compile C99?
- AstralStorm 10y agoPython is an interpreted language, so the comparison is meaningless. It is cheap to deal with the ambiguity in the compiler using compile options. You could go the Java way and have separate class path (e.g. In PyPy, as CPython 2.x is dead). But this does not allow you to mix and match code for older version anyway. Java was designed from start to allow this and it is why it's so stagnant.
- chongli 10y agoPython is an interpreted language, so the comparison is meaningless. Bash has no problem running old Bourne shell scripts from ancient history. Browsers have no problem running JavaScript from the mid-90s. I fail to see how being an interpreted language is a valid reason for breaking backwards compatibility.
- aangjie 10y agobeing interpreted is not a valid reason, but the string class for both bytes and text is a valid reason.
- kbenson 10y agoI used C as an example of what happens when you don't change, not what happens when you do. C is used decades later with little change because of momentum and history, not because it's the pinnacle of computer languages. C is not a good language. It was a good language, but that was a few decades ago. We've progressed past the point where C's shortcomings are excused by there not being alternatives that address those shortcomings while offering comparable features. At this point, we're just finally edging past of the local maxima created by C's exceptional popularity and the knock-on effects of that popularity, such as most operating systems (and all popular open source ones) being written in C.