Announcement

Collapse
No announcement yet.

Python 3.9 Released With Multi-Processing Improvements, New Parser

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Python 3.9 Released With Multi-Processing Improvements, New Parser

    Phoronix: Python 3.9 Released With Multi-Processing Improvements, New Parser

    Python 3.9 is out today as the newest feature update to this extremely popular language in open-source crowds...

    Phoronix, Linux Hardware Reviews, Linux hardware benchmarks, Linux server benchmarks, Linux benchmarking, Desktop Linux, Linux performance, Open Source graphics, Linux How To, Ubuntu benchmarks, Ubuntu hardware, Phoronix Test Suite

  • #2
    That's very exciting. Multiprocessing improvements won't change the spark driven high end, but for home users will be very handy.

    Comment


    • #3
      Originally posted by thelongdivider View Post
      That's very exciting. Multiprocessing improvements won't change the spark driven high end, but for home users will be very handy.
      In my experiences, multiprocessing goes down when you spark drive some high end

      Comment


      • #4
        The sensational "multiprocessing improvements" that made the headline amount to a minor addition to a queue API that happens to be in the 'multiprocessing' module. I understand you are performance-focused here, but that change is not even newsworthy.

        Comment


        • #5
          Originally posted by p91paul View Post
          The sensational "multiprocessing improvements" that made the headline amount to a minor addition to a queue API that happens to be in the 'multiprocessing' module. I understand you are performance-focused here, but that change is not even newsworthy.
          Indeed this is a total fuck-up to claim that this minor change is somehow news or a sizeable improvement to "multiprocessing". (Sorry for stripping those extraneous hyphens that Michael is obsessed with.)

          However, there seems to be another performance-related change that might bear some weight, that is the optimization of container types using the vectorcall protocol. I didn't dig around much, but one overly simplified benchmark using tuple showed improvements in the range of 20 to 40 %. (https://github.com/python/cpython/pull/18936)

          Comment


          • #6
            Originally posted by frank007
            In my opinion Python is dangerous. The last dangerous issue I remember: writing some characters not defined in a certain point of the program gave me an obvious error message; after substituting those characters with a simple space the program was executed instead of gaving me another error message, but the result of the execution was obviously wrong. I get very often a mysterious issue in using Python for making a gui program sooner or later. In my opinion Python is not good.
            PEBKAC

            Comment


            • #7
              I like Python because it comes preinstalled on Ubuntu and it is simple and easy to use.

              Really nice in version 3.9 that you can annotate without having to import typing, but I wish it always was like this, I don't know why they made it so you had to import typing and refer to List and Dict instead of as it is now where you refer to list and dict without have to bother with importing typing.

              Things I don't like about Python is that there async HTTP client in the standard library, and urllib sucks. The third-party Python package "requests" is great, but its unfortunate it or something like it is not included in the standard library.

              Another thing I don't like about Python is that if you take a break from Python then some months later it is easy to confuse static fields with instance fields on classes. Perhaps the properties are a bit confusing too, but maybe that is just to me.

              Comment


              • #8
                Originally posted by frank007
                In my opinion Python is dangerous. The last dangerous issue I remember: writing some characters not defined in a certain point of the program gave me an obvious error message; after substituting those characters with a simple space the program was executed instead of gaving me another error message, but the result of the execution was obviously wrong. I get very often a mysterious issue in using Python for making a gui program sooner or later. In my opinion Python is not good.
                I've no idea what you're on about but it's very likely you're just doing it wrong, or you don't understand the code you're writing vs what you're trying to accomplish.

                Comment


                • #9
                  Originally posted by frank007
                  In my opinion Python is dangerous. The last dangerous issue I remember: writing some characters not defined in a certain point of the program gave me an obvious error message; after substituting those characters with a simple space the program was executed instead of gaving me another error message, but the result of the execution was obviously wrong. I get very often a mysterious issue in using Python for making a gui program sooner or later. In my opinion Python is not good.
                  Yup, your obscure and totally not-your-fault problem is a compelling reason why Python as a whole is not a good language.

                  /s

                  Comment


                  • #10
                    Somewhat surprising. AFAIK PEG parsing is more error prone (unless one has an extensive test suite) and slower than traditional LALR(1). In PEG, a parse error won't necessarily result in an error message, instead a wrong parse tree is selected.

                    Comment

                    Working...
                    X