Nixpkgs: webkitgtk-2.20: build broken on hydra

[mdorman]

https://hydra.nixos.org/build/71902079/nixlog/1

[jtojnar]

It is an intermittent failure, see https://github.com/NixOS/nixpkgs/pull/36893, not much we can do about it other than restarting the build. I opened an upstream issue https://bugs.webkit.org/show_bug.cgi?id=183788

[mdorman]

I've now tried doing a _third_ build locally for this: all have failed. That doesn't seem intermittent to me.

[jtojnar]

It is succeeding for me most of the time, hydra also managed to build it in the past.

[matthewbauer]

Maybe try setting enableParallelBuilding = false; ? Not sure if it will help but it has helped with some derivations in the past.

[jtojnar]

@hedning managed to reproduce it even with parallel building disabled: https://github.com/NixOS/nixpkgs/pull/36893#issuecomment-373922664

[pbogdan]

Interestingly _lowering_ parallelism is what seems to trigger the issue for me. If I build with all cores (8) it seems to get past that part of the build all the time, I'm yet to see it succeed when building with a single core.

[hedning]

Can confirm that low parallelism is the problem. nix build -f . webkitgtk --option cores 4 is building successfully, which is the first time it has worked for me.

I'm yet to see it succeed when building with a single core.

That's consistent with what I've seen, not having it build successfully a single time (and it has always only used a single core).

[aristidb]

I also got the issue when doing a nixos-rebuild on my 2-core laptop, but it succeeded on the same laptop when using nix-build, which means that all cores are available for just webkitgtk.

This is the error message I got:

builder for '/nix/store/5a00rqavnjk83naq1h6mgmbdgc1v1f0r-webkitgtk-2.20.0.drv' failed with exit code 1; last 10 log lines:
                   from ../Source/JavaScriptCore/heap/Heap.h:33,
                   from ../Source/JavaScriptCore/heap/DeferGC.h:29,
                   from ../Source/JavaScriptCore/runtime/ConcurrentJSLock.h:28,
                   from ../Source/JavaScriptCore/bytecode/ArrayProfile.h:28,
                   from ../Source/JavaScriptCore/llint/LLIntOffsetsExtractor.cpp:28:
  ../Source/JavaScriptCore/API/JSHeapFinalizerPrivate.h:29:41: fatal error: JavaScriptCore/JSContextRef.h: No such file or directory
   #include <JavaScriptCore/JSContextRef.h>
                                           ^
  compilation terminated.
ninja: build stopped: subcommand failed.

[hedning]

Would it make sense to add the stopgap fix suggested in https://bugs.webkit.org/show_bug.cgi?id=183788#c4? So something like this:
```diff
diff --git a/pkgs/development/libraries/webkitgtk/2.20.nix b/pkgs/development/libraries/webkitgtk/2.20.nix
index 8b44f11ee93..81e9b9855da 100644
--- a/pkgs/development/libraries/webkitgtk/2.20.nix
+++ b/pkgs/development/libraries/webkitgtk/2.20.nix
@@ -60,6 +60,12 @@ stdenv.mkDerivation rec {
"-DENABLE_GTKDOC=OFF"
];

• postConfigure = ''
• # A stopgap for a non-deterministic build failure when using only one core
• # Upstream bug: https://bugs.webkit.org/show_bug.cgi?id=183788#c4
• ninja JavaScriptCoreForwardingHeaders WTFForwardingHeaders
• '';
  +
  NIX_CFLAGS_COMPILE = optionalString stdenv.isDarwin " -lintl";

[LinuxCao]

Open the Source/JavaScriptCore/CMakeLists.txt file
Add the following in line 258:
add_dependencies(LLIntOffsetsExtractor JavaScriptCoreForwardingHeaders)

[jtojnar]

This will be no longer neccesary with https://github.com/NixOS/nixpkgs/pull/45950