Actually, I opened this thread: https://groups.google.com/forum/?fromgroups=#!topic/julia-users/IjG2ERHVjz0 when I needed to revisit prolog and first order logic for my AI exam... I think Julia can be great for DSLs! The current downside of Julia being very young and not having any established IDE can turn into an upside, if we use this to make a deeper connection between the parser and the IDE. This way we could offer correct syntax highlighting and parsing for any user defined DSL, if they use the same interfaces. I imagine that Julia might become a nice hub for other languages, with its low overhead for calling functions from other languages. If other languages can be integrated as nicely as Keno has done it with his Cxx package, together with good IDE support, this could be huge. Deciding to use Julia wouldn't mean to decide against using other languages. (As much as I like to believe that Julia could be the one and only language, this is still very powerful)
2014-12-04 23:07 GMT+01:00 Páll Haraldsson <pall.haralds...@gmail.com>: > > > On Thursday, December 4, 2014 7:21:21 PM UTC, Isaiah wrote: >> >> Please: consider the advice given on the issue tracker and try to ask a >> small number of specific questions. Also, please search the issue tracker >> and mailing list (e.g., the question about assembly language has been >> discussed several times). >> > > Sorry about that, I went a little overboard with side-questions. About the > assembly language, I think it is not in, but it's probably least important > of everything I had to say. > > >> >> In general, asking questions like "what is the best language ever" is not >> going to lead to a very productive discussion. Programming languages are >> tools, not belief systems. Different tools are most appropriate for >> different purposes. >> > > I was trying not to ask what is the best language ever, but would like to > know if there are better available, at least for most uses. > > Yes, they are tools, I believe we have more than we need (thousands), much > more than the "different purposes". I can see why Julia is good (made) for > "technical" (e.g. speed) but it seems a good fit for lots of other uses > that use inferior languages. > >> >> In general, the answer to the question "should I use Julia in production" >> is right now going to be "no". Or more specifically: if it is necessary to >> ask, then the default answer is "no". There are certainly people using >> Julia "in production" already, for some definitions of "production" >> (possibly even with money on the line). But these people usually have the >> experience to evaluate the trade-offs, and the wherewithal to implement >> whatever they need. >> >> Julia lacks certain language features by choice. For example: it is not >> designed to write an OS kernel and is unlikely to be a good choice for hard >> realtime, microcontrollers, or other such uses in the near future >> > > I would think Julia could work for microcontrollers right now even (or > with little changes).. I was surprised to see Python proposed for such use > using modified but stil simple GC (not Pythons reference counting) and > still work for hard real-time. This is from memory. > > You would be crazy to start building a new OS from scratch. Only sane way > would be to use Linux and C (GCCism but now also compiles with LLVM) and > I'm not saying Linus would go with it, but Julia could be embedded in its C > code.. :) > > (maybe in a few years there will be cross-compilers for a subset of the >> language). Julia is not specifically designed for logic programming or >> theorem proving; certainly those things are possible given enough work, but >> there is probably a better language available (Coq or various Haskell >> flavors, for example). >> > > I was going to write a separate post on logic > programming/Prolog/declarative.. > > Prolog is mindblowing for what it's good for. I wouldn't touch it for > anything else. Racket supports Prolog (and ALGOL) using its macrosystem. It > feels like cheating if you say you support declarative programming with an > alternative syntax.. Couldn't Julia do the same? Or just embed a C Prolog > runtime? Or just implement the backtracking in a library and not use Prolog > syntax for Horn-clauses? Is declarative used really for anything else > (besides SQL)? > > >> >> In other cases, the lacking areas are due to prioritization and the still >> relatively-small resources available to fully implement desired features. >> For example, debugging, improved allocation and garbage collection (but, >> these are in progress!), as well as many library areas. >> > > "PL/1 (Programming Language number 1) was also a langage so much better > than anything previously invented" ok.. people I worked with used it in > 1996, was this a popular opinion when it first came out? > > Best regards, > Palli. >
