5 ms·

Julia v0.5.0 Released

114 points by dlss 10y ago

qd6pwu4 10y ago
Last time I used julia was in Juno, the loading of package was really slow, is this problem solved now? It once gave me a lot of hope...but the result was a little bit disappointing given that it claims so much. It has macro, if only it has lisp syntax...
- sundarurfriend 10y ago
  > the loading of package was really slow Was this on a Unix or on Windows? This has been my experience too on Windows, both the initial startup of Julia and the loading of packages was pretty slow (this was in 0.4.2 I believe). > It has macro, if only it has lisp syntax... Honestly, it's a language whose audience is at least 50% non-professional-programmers (scientists, engineers, statisticians), so with a lisp-y syntax it would probably have died a sputtering death. There's a lot of other things to love about the language though, whether superficial (being able to say `2y + 3` or `if a ≠ b && e ∈ S` if you wanted to), practical (pretty good FFIs to C, C++, Python, MATLAB and R already, decent-sized and quickly growing package repo), theoretical (mutliple dispatch, the type system).
  - qd6pwu4 10y ago
    > Was this on Unix or on Windows? It was on Mac OS. > Honestly... Yes, the language has quite a few lovely features, that's why it makes me both love and hate it...so I chose to be away for a while and see how it develops in the future.
  - stonemetal 10y ago
    To me it seems like it is professional programmers who are stodgy about their syntax. Non pros aren't as fussy when presented with something non standard because they haven't been married to the standard for 20 years.
    - eggy 10y ago
      I like to do math in C, J, Forth or Scheme which all have different syntax and a combination of prefix, postfix and infix. I would like a Lisp syntax too, and you can go into Lisp in Julia, since initially it used femtolisp written by Jeff Bezanson (one of the creators of Julia). I am not sure how much functionality is available from the Lisp prompt in the console. There's also a Lisp syntax package for it IIRC. I am waiting for it to stabilize more, and then I'll revisit it. I like now that I have the choice of zero-indexing, since I am used to that from mathematics, C, and J, although APL gives a switch to choose between 0 or 1 indexing.
  - joelhoffman 10y ago
    >with a lisp-y syntax it would probably have died a sputtering death Tell that to 30 years of Autocad users.
sundarurfriend 10y ago
Quite a milestone for Julia, and an important step forward towards stability as well: I believe a core Julia team member (Viral Shah IIRC) said that there would be only one or two major releases between 0.5 and 1.0. One thing I like about Julia development is how receptive they are to community ideas, while also making sure the suggestions fit into the larger pattern and rejecting those that don't. (The "Support for arrays with indexing starting at values different from 1" thing does worry me, since it has the potential to screw things up gloriously, but it seems[1] they predicted HN will gnaw at this one. :) And I'm yet to read the whole of that issue, maybe it'll end up convincing me it's a good idea.) [1] https://github.com/JuliaLang/julia/pull/16260 https://github.com/JuliaLang/julia/pull/16260
- mvardin 10y ago
  Given the long list of "breaking changes" and "Deprecated or removed" I think it is premature to suggest that they are moving towards stability.
  - cschmidt 10y ago
    I think the idea is that they're trying to make the breaking changes sooner rather than later. It should settle down after this release.
- mauro3 10y ago
  The outlook toward Julia 1.0 was given by Stefan Karpinski in his JuliaCon 2016 talk: https://youtu.be/5gXMpbY1kJY https://youtu.be/5gXMpbY1kJY . But yes, the plan is to have a 0.6 and then 1.0.
yla92 10y ago
Prior discussion: https://news.ycombinator.com/item?id=12535072 https://news.ycombinator.com/item?id=12535072
tempodox 10y ago
I take it, there is still no support for the production of stand-alone binary executables? Rumours persist that this feature is planned. Or are they just plain wrong?
- ihnorton 10y ago
  There has been ongoing work toward supporting this feature. See: http://juliacomputing.com/blog/2016/03/10/j2c-announcement.html http://juliacomputing.com/blog/2016/03/10/j2c-announcement.h... http://juliacomputing.com/blog/2016/02/09/static-julia.html http://juliacomputing.com/blog/2016/02/09/static-julia.html
  - whyrt12 10y ago
    How close are we to web assembly compilation?
    - ihnorton 10y ago
      I'm not aware of anyone working on it. There were some attempts with Emscripten which ran in to unsupported instructions; some of those may be supported by newer versions of Emscripten, but I haven't heard of any attempts to get this going recently.
      - whyrt12 10y ago
        Gotcha. Is it definitely technically feasible, or is that still unknown?
        ihnorton 10y ago
        For Julia core it should be feasible, but possibly requiring patches to Emscripten, so the level of effort is still unknown. The other issue is the C and Fortran numerical libraries -- some of which use inline-assembly, others (Fortran) don't have a working LLVM frontend as far as I know.
boromi 10y ago
duplicate of: https://news.ycombinator.com/item?id=12535072 https://news.ycombinator.com/item?id=12535072

114 points by dlss 10y ago

Last time I used julia was in Juno, the loading of package was really slow, is this problem solved now? It once gave me a lot of hope...but the result was a little bit disappointing given that it claims so much. It has macro, if only it has lisp syntax...

sundarurfriend 10y ago
> the loading of package was really slow Was this on a Unix or on Windows? This has been my experience too on Windows, both the initial startup of Julia and the loading of packages was pretty slow (this was in 0.4.2 I believe). > It has macro, if only it has lisp syntax... Honestly, it's a language whose audience is at least 50% non-professional-programmers (scientists, engineers, statisticians), so with a lisp-y syntax it would probably have died a sputtering death. There's a lot of other things to love about the language though, whether superficial (being able to say `2y + 3` or `if a ≠ b && e ∈ S` if you wanted to), practical (pretty good FFIs to C, C++, Python, MATLAB and R already, decent-sized and quickly growing package repo), theoretical (mutliple dispatch, the type system).
- qd6pwu4 10y ago
  > Was this on Unix or on Windows? It was on Mac OS. > Honestly... Yes, the language has quite a few lovely features, that's why it makes me both love and hate it...so I chose to be away for a while and see how it develops in the future.
- stonemetal 10y ago
  To me it seems like it is professional programmers who are stodgy about their syntax. Non pros aren't as fussy when presented with something non standard because they haven't been married to the standard for 20 years.
  - eggy 10y ago
    I like to do math in C, J, Forth or Scheme which all have different syntax and a combination of prefix, postfix and infix. I would like a Lisp syntax too, and you can go into Lisp in Julia, since initially it used femtolisp written by Jeff Bezanson (one of the creators of Julia). I am not sure how much functionality is available from the Lisp prompt in the console. There's also a Lisp syntax package for it IIRC. I am waiting for it to stabilize more, and then I'll revisit it. I like now that I have the choice of zero-indexing, since I am used to that from mathematics, C, and J, although APL gives a switch to choose between 0 or 1 indexing.
- joelhoffman 10y ago
  >with a lisp-y syntax it would probably have died a sputtering death Tell that to 30 years of Autocad users.

sundarurfriend 10y ago

Quite a milestone for Julia, and an important step forward towards stability as well: I believe a core Julia team member (Viral Shah IIRC) said that there would be only one or two major releases between 0.5 and 1.0. One thing I like about Julia development is how receptive they are to community ideas, while also making sure the suggestions fit into the larger pattern and rejecting those that don't. (The "Support for arrays with indexing starting at values different from 1" thing does worry me, since it has the potential to screw things up gloriously, but it seems[1] they predicted HN will gnaw at this one. :) And I'm yet to read the whole of that issue, maybe it'll end up convincing me it's a good idea.) [1] https://github.com/JuliaLang/julia/pull/16260 https://github.com/JuliaLang/julia/pull/16260

mvardin 10y ago
Given the long list of "breaking changes" and "Deprecated or removed" I think it is premature to suggest that they are moving towards stability.
- cschmidt 10y ago
  I think the idea is that they're trying to make the breaking changes sooner rather than later. It should settle down after this release.
mauro3 10y ago
The outlook toward Julia 1.0 was given by Stefan Karpinski in his JuliaCon 2016 talk: https://youtu.be/5gXMpbY1kJY https://youtu.be/5gXMpbY1kJY . But yes, the plan is to have a 0.6 and then 1.0.

yla92 10y ago

Prior discussion: https://news.ycombinator.com/item?id=12535072 https://news.ycombinator.com/item?id=12535072

tempodox 10y ago

I take it, there is still no support for the production of stand-alone binary executables? Rumours persist that this feature is planned. Or are they just plain wrong?

ihnorton 10y ago
There has been ongoing work toward supporting this feature. See: http://juliacomputing.com/blog/2016/03/10/j2c-announcement.html http://juliacomputing.com/blog/2016/03/10/j2c-announcement.h... http://juliacomputing.com/blog/2016/02/09/static-julia.html http://juliacomputing.com/blog/2016/02/09/static-julia.html
- whyrt12 10y ago
  How close are we to web assembly compilation?
  - ihnorton 10y ago
    I'm not aware of anyone working on it. There were some attempts with Emscripten which ran in to unsupported instructions; some of those may be supported by newer versions of Emscripten, but I haven't heard of any attempts to get this going recently.
    - whyrt12 10y ago
      Gotcha. Is it definitely technically feasible, or is that still unknown?
      - ihnorton 10y ago
        For Julia core it should be feasible, but possibly requiring patches to Emscripten, so the level of effort is still unknown. The other issue is the C and Fortran numerical libraries -- some of which use inline-assembly, others (Fortran) don't have a working LLVM frontend as far as I know.

boromi 10y ago

duplicate of: https://news.ycombinator.com/item?id=12535072 https://news.ycombinator.com/item?id=12535072