It's "mostly" working fine here (tested with "latest"):- Linux and OSX work out of the box (maybe because I have "accidentally" Java installed.- on Windows running the Closure pass fails because of missing JavaWill Java be provided by emsdk or do I need to care of that myself?
And some "nice-to-have" feedback:- Is it already time to write tickets for the latest-releases-upstream version? Compilation with my code appears to work, but I'm getting runtime errors for EM_JS functions:failed to asynchronously prepare wasm: LinkError: WebAssembly.instantiate(): Import #61 module="env" function="_sapp_js_create_textfield" error: function import requires a callableclear-sapp.html:39 LinkError: WebAssembly.instantiate(): Import #61 module="env" function="_sapp_js_create_textfield" error: function import requires a callable
- The very first emcc invocation after another SDK version is installed/activated (in my case during the cmake compiler detection) takes *very* long on Windows, something like 5..10 minutes, and without feedback, so it appears like the build process hangs. Under the hood there's some compilation happening (maybe binaryen?). The time this takes is so "uncomfortably long" that I would consider it a bug. May be in the future it would make sense to also provide precompiled artifacts for those hidden compilations for the "prebuilt" SDK versions?
Cheers,-Floh.On Wednesday, 29 May 2019 19:45:45 UTC+2, Alon Zakai wrote:If you get emsdk master, and doemsdk install latestemsdk activate latestthen it will install 1.38.33 (or later, if there is one by the time you read this) from the new emscripten-releases build infrastructure. We've done some early testing recently as mentioned here:After fixing some issues, there seem to be no showstoppers, so this is us moving to the next stage in the transition from the old builders to the new ones. (If all goes well then the stage after this will be to remove those old builders.)Please report any issues you see!Known issue: "no version information" warning on some linux machines, https://github.com/emscripten-core/emscripten/issues/8697, that we need to figure out.- Alon
--
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 emscripten-disc...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/emscripten-discuss/3ee49945-abce-4a0e-8814-69e38a1f1d49%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
Did it not say what it was doing during that pause? Like it should say if it was building binaryen, or system libraries, etc. It should print that even without EMCC_DEBUG=1 in the env, but that may provide even more info.
5-10 minutes sounds extremely long for either binaryen or system libraries, though. Was that on a slow machine? If not, then that sounds bad. Perhaps python doesn't detect how many cores there are and the builds are single-core, or maybe somehow our build of LLVM is debug and not release, or something like that?
Did it not say what it was doing during that pause? Like it should say if it was building binaryen, or system libraries, etc. It should print that even without EMCC_DEBUG=1 in the env, but that may provide even more info.No output, it happens during cmake's compiler / compile feature detection. Maybe cmake is supressing console output on purpose there to not spam the console with errors that might be expected during the detection.
5-10 minutes sounds extremely long for either binaryen or system libraries, though. Was that on a slow machine? If not, then that sounds bad. Perhaps python doesn't detect how many cores there are and the builds are single-core, or maybe somehow our build of LLVM is debug and not release, or something like that?It's not a particularly slow laptop, I have set it up for Win10 / Ubuntu dual boot, and the same process on Linux is about the same as on my Mac laptop (which is acceptable). I think it's a Windows issue, IME clang on Windows is a lot slower than on OSX or Linux for some reason.Maybe the Windows build is accidentally using a debug build of LLVM, I didn't pay attention to this, and it was downloading a prebuilt package, so that would be hard to notice anyway.Cheers,-Floh.
--
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 emscripten-disc...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/emscripten-discuss/56218175-49c2-4199-a291-39296350452a%40googlegroups.com.
Would be interesting to see in a fresh install what happens when building a hello world on windows, where the output isn't hidden.
Definitely file bugs on upstream if you find any! Yes, please file this one with a testcase, it doesn't look like a known issue.
--
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 emscripten-disc...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/emscripten-discuss/c678df7b-5330-4170-8e65-11c35a1d96b2%40googlegroups.com.
About compiling libc which takes so long - is compiling your own code slower as well?
On Thu, May 30, 2019 at 10:54 AM Floh <flo...@gmail.com> wrote:
--Would be interesting to see in a fresh install what happens when building a hello world on windows, where the output isn't hidden.It's libc which takes so long:cache:INFO: generating system library: libc.bc... (this will be cached in...
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 emscripten-discuss+unsub...@googlegroups.com.
> About compiling libc which takes so long - is compiling your own code slower as well?Yes, compiling my own code is also slower, but I've seen this since forever for the same stuff compiled via clang on OSX/Linux vs Windows. I just started to accept it as some weird clang oddity ;)
I haven't installed / activated emscripten SDKs so often in the past on Windows though, so I didn't notice this as much as now.
On Friday, 31 May 2019 04:35:01 UTC+2, Alon Zakai wrote:
About compiling libc which takes so long - is compiling your own code slower as well?
On Thu, May 30, 2019 at 10:54 AM Floh <flo...@gmail.com> wrote:
--Would be interesting to see in a fresh install what happens when building a hello world on windows, where the output isn't hidden.It's libc which takes so long:cache:INFO: generating system library: libc.bc... (this will be cached in...
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 emscripten-disc...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/emscripten-discuss/c678df7b-5330-4170-8e65-11c35a1d96b2%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
--
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 emscripten-disc...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/emscripten-discuss/e2632bd3-67cf-4364-b6c1-424faaae8be3%40googlegroups.com.
Thanks for the info, interesting.I doublechecked and we do build LLVM in release mode, on Windows and elsewhere. We do build with assertions on all platforms, which is something we can perhaps remove for speed, but it is useful.Another possible issue is that we use Python multiprocess code to compile libraries (so we use all cores), and I've heard it's slower on windows for various reasons. Perhaps there is something faster? I opened https://github.com/emscripten-core/emscripten/issues/8719- Alon
On Fri, May 31, 2019 at 12:39 AM Floh <flo...@gmail.com> wrote:
> About compiling libc which takes so long - is compiling your own code slower as well?Yes, compiling my own code is also slower, but I've seen this since forever for the same stuff compiled via clang on OSX/Linux vs Windows. I just started to accept it as some weird clang oddity ;)
I haven't installed / activated emscripten SDKs so often in the past on Windows though, so I didn't notice this as much as now.
On Friday, 31 May 2019 04:35:01 UTC+2, Alon Zakai wrote:
About compiling libc which takes so long - is compiling your own code slower as well?
On Thu, May 30, 2019 at 10:54 AM Floh <flo...@gmail.com> wrote:
--Would be interesting to see in a fresh install what happens when building a hello world on windows, where the output isn't hidden.It's libc which takes so long:cache:INFO: generating system library: libc.bc... (this will be cached in...
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 emscripten-discuss+unsub...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/emscripten-discuss/c678df7b-5330-4170-8e65-11c35a1d96b2%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
--
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 emscripten-discuss+unsub...@googlegroups.com.
It look like things have changed a bit recently.
Actually, you can know the unique identifier of the last
precompiled SDK available in the file
"emscripten-releases-tags.txt" file (after issuing "emsdk
update-tags" commands)
On my side the contents is:
{
"latest": "1.38.33",
"releases": {
"1.38.33": "3b8cff670e9233a6623563add831647e8689a86b" //
this is the unique identifier
}
}
Then with that info I can do:
./emsdk install sdk-releases-upstream-<the_unique_identifier>-64bit
This will install the 1.38.33. This is not very convenient I
think, but it works
Gabriel
--
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 emscripten-disc...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/emscripten-discuss/04241de6-3244-4c6f-b9ce-064d790815c6%40googlegroups.com.
To unsubscribe from this group and stop receiving emails from it, send an email to emscripten-discuss+unsub...@googlegroups.com.
--
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 emscripten-disc...@googlegroups.com.
Note that I am not sure if we have to use
sdk-releases-upstream-<the_unique_identifier>-64bit
or
sdk-releases-fastcomp-<the_unique_identifier>-64bit
I don't get exactly what is the difference between the two
To unsubscribe from this group and stop receiving emails from it, send an email to emscripten-disc...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/emscripten-discuss/45a0a10c-35ce-4ac8-bb3f-8e9627bf0390%40googlegroups.com.
To unsubscribe from this group and stop receiving emails from it, send an email to emscripten-disc...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/emscripten-discuss/04241de6-3244-4c6f-b9ce-064d790815c6%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
--
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 emscripten-disc...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/emscripten-discuss/45a0a10c-35ce-4ac8-bb3f-8e9627bf0390%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
--
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 emscripten-disc...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/emscripten-discuss/48a943aa-9768-6337-aa9a-1e22f3262265%40gmail.com.
You received this message because you are subscribed to a topic in the Google Groups "emscripten-discuss" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/emscripten-discuss/fo0W09OYYfk/unsubscribe.
To unsubscribe from this group and all its topics, send an email to emscripten-disc...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/emscripten-discuss/CAEX4NpRnyqZn7jjmTY2aPZNcYqpb3fSSomtGdRUO6vc-rj2Hgw%40mail.gmail.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/emscripten-discuss/D139BD38-8652-4FAD-A7E7-7C3D4118773D%40pspdfkit.com.
Ok, good to know about the upcoming LLVM backend !
I did a few tests of that new backend with the Doom 3 port.
Except for EMTEPRETER which is not supported (oops, too bad. I can
no more yield to the browser during the long loading screens!),
everything else run fine.
I did not noticed significant runtime performance improvement
though, but I guess it is because the real bottleneck in D3wasm
lies in the GL=>WebGL glue code (or somewhere else, but not in
the "pure" wasm binary).
It look like the compilation/link is faster, but maybe that's me.
To view this discussion on the web visit https://groups.google.com/d/msgid/emscripten-discuss/CAEX4NpRnyqZn7jjmTY2aPZNcYqpb3fSSomtGdRUO6vc-rj2Hgw%40mail.gmail.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/emscripten-discuss/beae7d48-bc9b-ba27-0966-7a30d77868f2%40gmail.com.
Oops, sorry, there is issues in fact with the new LLVM backend ! (NB: I made a mistake by compiling/running the fastcomp version of D3, and not the LLVM version).
So the error happens at link stage, with the following output:
error: undefined symbol: g$SIMDProcessor
error: undefined symbol:
g$_ZN11idHashIndex13INVALID_INDEXE
error: undefined symbol: g$_ZN5idLib10cvarSystemE
error: undefined symbol: g$_ZN5idLib10fileSystemE
error: undefined symbol: g$_ZN5idLib3sysE
error: undefined symbol: g$_ZN5idLib6commonE
error: undefined symbol: g$_ZN6idCVar10staticVarsE
error: undefined symbol: g$_ZN6idMatX7tempPtrE
error: undefined symbol: g$_ZN6idMath10SQRT_THREEE
error: undefined symbol: g$_ZN6idMath11FLT_EPSILONE
error: undefined symbol: g$_ZN6idMath11SQRT_1OVER2E
error: undefined symbol: g$_ZN6idMath12ONEFOURTH_PIE
error: undefined symbol: g$_ZN6idMath2PIE
error: undefined symbol: g$_ZN6idMath5iSqrtE
error: undefined symbol: g$_ZN6idMath6TWO_PIE
error: undefined symbol: g$_ZN6idMath7HALF_PIE
error: undefined symbol: g$_ZN6idMath8INFINITYE
error: undefined symbol: g$_ZN6idMath9M_DEG2RADE
error: undefined symbol: g$_ZN6idMath9M_RAD2DEGE
error: undefined symbol: g$_ZN6idVecX7tempPtrE
error: undefined symbol: g$_ZN6idVecX9tempIndexE
error: undefined symbol: g$_ZTV14idSIMD_Generic
error: undefined symbol: g$common
error: undefined symbol: g$cvarSystem
error: undefined symbol: g$mat3_identity
error: undefined symbol: g$vec3_origin
error: undefined symbol: __memory_base
error: undefined symbol: __table_base
Most of the undefined symbols are global variables or constants
in D3 code. I am not sure what is the root cause of the problem.
However, "__table_base" and "__memory_base" undefined symbols
looks quite suspect in my opinion. Any hints about where to look ?
To view this discussion on the web visit https://groups.google.com/d/msgid/emscripten-discuss/652ae7a5-4405-2d77-2ceb-a5d208a9dd33%40gmail.com.