On 1/9/2017 11:07 PM, BGB wrote:
On 1/9/2017 10:06 PM, Charles Vaughn wrote:
optimizer.exe is run from multiple workers in a chunked fashion. I
would recommend you build emscripten from scratch so you can take
advantage of the latest version.
I would also recommend Ninja over make, but you'll need to switch to
a CMake project, though that has the benefit of being able to
generate both ninja projects and Visual Studio ones from the same
source.
the Windows 'emsdk' tool seems to list 1.35.0 as latest.
trying to build the Linux version in WS4L, which has 1.37.1 available.
had to manually upgrade CMake, as the "latest" version it had was IIRC
2.8.1, but LLVM seemed to want a minimum of 3.4.3...
downloaded CMake, now building with 3.7.1, seems to be working, and at
least thus far it doesn't seem I am having to do the usual thing of
chasing down and installing dependencies. the "dependency dance" is
IMO probably one of the main annoyances of trying to rebuild most FOSS
stuff from source.
trying to build project with newly built emcc:
<---
asm2wasm:
/home/bgb/.emscripten_ports/binaryen/binaryen-version_21/src/asm2wasm.h:2066:
wasm::Asm2WasmBuilder::processFunction(cashew::Ref)::__lambda15:
Assertion `index >= 0' failed.
Traceback (most recent call last):
File "/home/bgb/emsdk_portable/emscripten/master/emcc", line 13, in
<module>
emcc.run()
File "/home/bgb/emsdk_portable/emscripten/master/emcc.py", line
2099, in run
subprocess.check_call(cmd, stdout=open(wasm_text_target, 'w'))
File "/usr/lib/python2.7/subprocess.py", line 540, in check_call
raise CalledProcessError(retcode, cmd)
subprocess.CalledProcessError: Command
'[u'/home/bgb/.emscripten_ports/binaryen/binaryen-version_21/bin/asm2wasm',
'deeptail0.asm.js', '--total-memory=536870912', '-O3',
'--mem-init=deeptail0.html.mem', '--mem-base=1024', '--wasm-only']'
returned non-zero exit status -6
--->
( this appears to have to do with compiling switch() statements or
similar... )
ADD:
it still works if I don't use WASM though, but I would like to be able
to use WASM.
a quick skim of the code didn't see anything obviously amiss, so don't
know how to fix it on my end (not entirely sure how that assert can
fail, logic implies it probably shouldn't fail).
( though, FWIW, the codebase does have a few relatively large switch
statements in it, ... )
as for build speeds (with WASM disabled):
"make": 2 minutes;
"make -j": 36 seconds.
I will regard this as an improvement (so, normal/full = 10 min -> 2 min;
make -j = 2 min -> 36 sec).
quick tests, it appears a lot of the OpenGL related issues still exist
though (at least the more obvious ones).
--
You received this message because you are subscribed to the Google Groups
"emscripten-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/d/optout.
