The contrast between "flexibility of choice" and "tyranny of one standard" is, um, a little over-drawn. I have three different Common Lisp systems on this laptop: CMUCL, SBCL, and CCL. I have the flexibility of choice between them *because* the adhere to a common standard. Each of them has its own extra bits, but I am free to choose between them because everything *else* is not idiosyncratic. I have a choice of machine learning libraries for Python, but that is useful to me because the *rest* of Python isn't different, so there's a huge amount of stuff I can use with an ML program.
In Smalltalk I cannot even port a program that opens a file for input, another one for output, and copies the one to the other, despite everything I need being in the ANSI Smalltalk standard. I mean come ON! I have two commercial Smalltalks, two formerly commercial Smalltalks, three free Smalltalks, and several Smalltalk implementations that were never quite finished, and no two of them handle Unicode the same way! Think about it: you have "Καλημέρα κόσμε" in a file, and no part of trying to copy that file is portable. Let's not mention Sockets, OK? Let's just not go there. Choice and consistency are NOT a dichotomy. They are both important. You need consistency at one level to construct choices at the next level. I note that there are type systems available for Prolog, Erlang, Scheme, and Python (https://realpython.com/python-type-checking/) and I've done work on a type inference program for AWK. Also that there are dynamic type facilities for Haskell and Clean. So dynamic types -vs- static types are not a hard dichotomy either. On Tue, 11 Aug 2020 at 13:08, horrido <horrido.hobb...@gmail.com> wrote: > It's true, Smalltalk faces the same dilemma as Linux and Lisp. As a > /family/ > of languages, portability is a genuine issue. > > There's no getting around this dichotomy. You can have either a flexibility > of choice or the tyranny of one standard, but not both. > > The decision is a fact of life that we face frequently. You can have either > the flexibility of dynamic typing or the safety of static typing, but not > both. You can have either the natural modelling of the real world due to > state mutation or the mathematical safety of immutability, but not both. > You > can have either the portability of a virtual machine or the > close-to-the-metal performance of native code generation, but not both (JIT > compilation notwithstanding). > > Life is about choices. There will always be a place for different > technologies. Smalltalk will not always be the ideal choice. That's why > there are five entirely different languages that dominate our industry > (Python, JavaScript, Java, C#, C++). > > There is no reason why there can't be a sixth, especially if it can > dramatically improve our productivity and make programming less cognitively > stressful. Surely, that's worth fighting for. > > Enough? > > > > Richard O'Keefe wrote > > Here is a challenge: What is "Smalltalk"? > > VAST, VW, and Pharo are quite different environments. > > To the extent that they share a common syntax (which > > they don't, quite), fine, but porting nontrivial > > code between them is NOT easy. They certainly have > > very little in common as GUI kits. All praise and > > thanks to the people who *have* ported stuff. > > > > > > On Tue, 11 Aug 2020 at 01:05, Richard Kenneth Eng < > > > horrido.hobbies@ > > > > > > wrote: > > > >> https://smalltalk.tech.blog/2020/08/10/smalltalks-successor/ > >> > >> A bold claim. It'll be interesting to see if anybody challenges me on > >> this. > >> > >> > > > > > > -- > Sent from: http://forum.world.st/Pharo-Smalltalk-Users-f1310670.html > >
