Announcement

Collapse
No announcement yet.

Meson 0.59 Build System Adds First Class Cython, Wine Resource Compiler Support

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

  • Meson 0.59 Build System Adds First Class Cython, Wine Resource Compiler Support

    Phoronix: Meson 0.59 Build System Adds First Class Cython, Wine Resource Compiler Support

    The open-source Meson build system that continues to be increasingly used by open-source projects and other software is out with version 0.59, which continues tacking on more features...

    https://www.phoronix.com/scan.php?pa...9-Build-System

  • #2
    How likely is the new wine feature going to be used for dxvk, vkd3d ..proton ?
    Last edited by CochainComplex; 19 July 2021, 01:08 PM.

    Comment


    • #3
      Last time I checked, Meson had serious issues with compilers it doesn't know about, even when forced through exporting as environment variable (e.g. the new Intel compiler which is based on Clang didn't work at all). Using Polly as an external plugin (as can be found on some distro tool chains, e.g. Suse or Arch) is also not possible, don't ask me why. Is it too much to ask of a build system to be more flexible?!

      Comment


      • #4
        Originally posted by ms178 View Post
        Last time I checked, Meson had serious issues with compilers it doesn't know about, even when forced through exporting as environment variable (e.g. the new Intel compiler which is based on Clang didn't work at all). Using Polly as an external plugin (as can be found on some distro tool chains, e.g. Suse or Arch) is also not possible, don't ask me why. Is it too much to ask of a build system to be more flexible?!
        The entire design of Meson was a tradeoff to centralize all the complicated logic inside Meson so that it's easier for devs to use in their projects. That's necessarily going to mean less flexibility.

        That means if you want to use brand new stuff that Meson doesn't understand, it's going to be a few months before someone makes it work. If that sounds bad, then Meson probably isn't the build system you want to use.

        Comment


        • #5
          Originally posted by smitty3268 View Post

          The entire design of Meson was a tradeoff to centralize all the complicated logic inside Meson so that it's easier for devs to use in their projects. That's necessarily going to mean less flexibility.
          This limits me to use compilers Meson knows about and even with LLVM/Clang, not all functionality is usable depending on how your distro compiled LLVM/Clang. Trying to convince distros like Arch and Suse to build and package LLVM/Clang/Polly differently would not be the faster solution to that part of my mentioned problems and building my own toolchain is not that easy on some of these distros and very time consuming.

          Originally posted by smitty3268 View Post
          That means if you want to use brand new stuff that Meson doesn't understand, it's going to be a few months before someone makes it work. If that sounds bad, then Meson probably isn't the build system you want to use.
          My Github issues are open long enough in their issue tracker with no fix in sight. I had far fewer problems trying to compile Mesa with Autotools than Meson in this regard. And it is not just me complaining about these open issues. If they were fixed in a couple of weeks, no one would mind, but it seems that the devs are overloaded with other tasks.

          Comment


          • #6
            Originally posted by ms178 View Post

            This limits me to use compilers Meson knows about and even with LLVM/Clang, not all functionality is usable depending on how your distro compiled LLVM/Clang. Trying to convince distros like Arch and Suse to build and package LLVM/Clang/Polly differently would not be the faster solution to that part of my mentioned problems and building my own toolchain is not that easy on some of these distros and very time consuming.



            My Github issues are open long enough in their issue tracker with no fix in sight. I had far fewer problems trying to compile Mesa with Autotools than Meson in this regard. And it is not just me complaining about these open issues. If they were fixed in a couple of weeks, no one would mind, but it seems that the devs are overloaded with other tasks.
            Upstream developer here, It's an open source project, and I've offered to review any patches you want to write for enabling new compilers. Several of people have contributed to getting their niché compiler working, and at this point with the abstractions we have in place it's not that hard. We're mostly volunteers, and even those of us who can use work time on Meson (I get to spend some work time on it), need to be able to justify that the work we're doing benefits our employer (or our project for our employer, as may be the case).

            Comment


            • #7
              Originally posted by dcbaker View Post

              Upstream developer here, It's an open source project, and I've offered to review any patches you want to write for enabling new compilers. Several of people have contributed to getting their niché compiler working, and at this point with the abstractions we have in place it's not that hard. We're mostly volunteers, and even those of us who can use work time on Meson (I get to spend some work time on it), need to be able to justify that the work we're doing benefits our employer (or our project for our employer, as may be the case).
              Hey dcbaker! This is not the first place where we have talked about all of my mentioned issues and I thank you for taking your time to look at all of them (the Polly issue for reference: https://gitlab.freedesktop.org/mesa/mesa/-/issues/4488). I am not a software programmer, I deal with the law in my profession, but I would love to contribute my testing experiences and bug reports to make Meson and other open source projects better in my spare time - it is just sad to see for me that my issues don't seem to be of any priority and staled for more than 8 months. I've noticed other people's issues are piling up on the Meson bug tracker, too. I cannot say if these numbers are concerning or not, but seeing the ratio of open vs. closed issues makes me wonder if the Meson project is still in a healthy state. By the way, I wouldn't call ICX/ICPX a niché compiler, after all it is the successor to ICC/ICPC and if Intel (your employer, albeit it's a different team's responsibility) wants me to use it (which I'd love to try out) it would be really useful to coordinate with the oneAPI team and build support for it in Meson, as a lot of open source projects use Meson nowadays. Frankly, I think that is not too much to ask from a multi-billion dollar corporation which wants me to use their hot new compiler.
              Last edited by ms178; 23 July 2021, 01:54 AM.

              Comment

              Working...
              X