"Extraordinary claims require extraordinary proof and the onus is on Lisp advocates to explain why there are so few real-world Lisp success stories. Calling everybody else stupid isn't good enough."
You seem to be suggesting that some technical shortcoming of Lisp held it back in the real world, while C and C++ had technical advantages that led to their success. I think the history of Lisp and the history of C tell a much different story.
While C was becoming the language of choice for OSes that ran on low-end computers (especially Unix), Lisp was confined to very expensive computers that were being targeted at a market that ultimately failed to materialize. C became popular by riding on the coattails of Unix, not because it is a great language or because it could compete with other languages on technical features, and C++ road on the coattails of C (maintaining some amount of compatibility etc.). Lisp, meanwhile, was held back by bad business decisions; no amount of technical superiority could have saved Lisp from the AI winter. By the time you could really talk about Lisp running on PCs or cheap hardware, it was too late: C was already widely used and there was already an enormous volume of C code that people were working with.
It is interesting that you mention ITA's use of C++. The big reason for that was the poor scalability of CMUCL's garbage collector, which they did not use at all, opting instead for memory-mapped arrays (hence C++, which was used to handle system calls and deal with pointers). Had ITA been developing for an OS written in Lisp (not unprecedented), C++ would never have entered the picture; but ITA was developing for an OS that exposed a C API and a C ABI, and using C or C++ to deal with system calls simply made the most sense. I suspect that if they were to try again today, they would use far less C++, if any; today's Lisp compilers come with (non-portable) extensions for POSIX system calls and for dealing with pointers, as well as disabling the garbage collector globally or on specific objects. It is not that Lisp itself was slow; just a specific feature of the Lisp systems they were using, and poor support for disabling or avoiding the use of that feature.
So which is more likely: that C and C++ are the best languages ever developed for the kinds of programs they are used for, or that they were just in the right place at the right time?
So which is more likely: that C and C++ are the best languages ever developed for the kinds of programs they are used for, or that they were just in the right place at the right time?
Or, they're better than any alternative for certain domains and the people that choose C++ know about Lisp, Haskell, Scala, Ocaml etc and still decide to use it despite its shortcomings?
It's a simple question - if Lisp really is superior for these kinds of apps then why isn't there some small, nimble team out there kicking ass with it?
HN is interesting because of the community and Y combinator's backing. The software behind it is laughably crude compared to other web forums. It could easily be replicated in any number of other languages, with better results. Maybe we'd even see the last of those "link expired" errors?
Surely you know the history to which the previous poster is referring? YC exists because pg and others build a company in Lisp and sold it for millions. They were a small team so they chose Lisp exactly because it helped them stay nimble.
They chose Lisp because they were comfortable with it - nimble is a social trait, not a language feature.
It's not like Lisp automatically makes everything successful - the market-trailing HN performance and UI iteration times are a strong argument that technical factors frequently do not determine success.
You seem to be suggesting that some technical shortcoming of Lisp held it back in the real world, while C and C++ had technical advantages that led to their success. I think the history of Lisp and the history of C tell a much different story.
While C was becoming the language of choice for OSes that ran on low-end computers (especially Unix), Lisp was confined to very expensive computers that were being targeted at a market that ultimately failed to materialize. C became popular by riding on the coattails of Unix, not because it is a great language or because it could compete with other languages on technical features, and C++ road on the coattails of C (maintaining some amount of compatibility etc.). Lisp, meanwhile, was held back by bad business decisions; no amount of technical superiority could have saved Lisp from the AI winter. By the time you could really talk about Lisp running on PCs or cheap hardware, it was too late: C was already widely used and there was already an enormous volume of C code that people were working with.
It is interesting that you mention ITA's use of C++. The big reason for that was the poor scalability of CMUCL's garbage collector, which they did not use at all, opting instead for memory-mapped arrays (hence C++, which was used to handle system calls and deal with pointers). Had ITA been developing for an OS written in Lisp (not unprecedented), C++ would never have entered the picture; but ITA was developing for an OS that exposed a C API and a C ABI, and using C or C++ to deal with system calls simply made the most sense. I suspect that if they were to try again today, they would use far less C++, if any; today's Lisp compilers come with (non-portable) extensions for POSIX system calls and for dealing with pointers, as well as disabling the garbage collector globally or on specific objects. It is not that Lisp itself was slow; just a specific feature of the Lisp systems they were using, and poor support for disabling or avoiding the use of that feature.
So which is more likely: that C and C++ are the best languages ever developed for the kinds of programs they are used for, or that they were just in the right place at the right time?