Ok, after mucking around for most of the day with cleaning up my build system integration I can say: "it works" (the "latest-releases-fastcomp" SDK version). I was expecting that the closure pass breaks, but looks like it works since I had a Java RTE "accidently" installed). I have only tested my sokol-samples so far, but these are quite "representative" I think.
I wrote down some feedback in this PR: https://github.com/emscripten-core/emsdk/pull/248 Cheers, -Floh. On Friday, 24 May 2019 23:44:38 UTC+2, Alon Zakai wrote: > > Yeah, that might be useful - I opened > https://github.com/emscripten-core/emsdk/pull/247 for further discussion. > I also wonder if a fixed dir might be better actually, also mentioned there > as a question. > > - Alon > > > On Fri, May 24, 2019 at 10:26 AM Floh <[email protected] <javascript:>> > wrote: > >> > However, when you do the "activate" step it populates the .emscripten >> file with the relevant paths - can you get it from there for your build >> system? >> >> Yes, I guess I can figure something out along those lines :) >> >> I'm planning to rework my emscripten integration so that it plays better >> with the emsdk (e.g. I'm currently not supporting easily switching between >> SDK versions). >> >> One thing that would be nice would be quering the current SDK path a'la >> the Xcode command line tools xcode-select and xcrun. >> >> The xcode-select tool is a bit similar to "emsdk activate", I can make >> one of many installed macOS/iOS/tvOS... SDKs the active one for the other >> command line tools... >> >> And xcrun has an option to show the currently active SDK path, for >> instance: >> >> > xcrun --show-sdk-path >> >> /Applications/Xcode.app/Contents/Developer/Platforms/MacOSX.platform/Developer/SDKs/MacOSX10.14.sdk >> >> ...and the output is in an "automation-friendly format", just the path >> and nothing else. >> >> So something like this would be nice: >> >> > emsdk show-sdk-path >> >> /Users/floh/.../fastcomp/3b8cff670e9233a6623563add831647e8689a86b/emscripten/ >> >> On Friday, 24 May 2019 19:13:34 UTC+2, Alon Zakai wrote: >>> >>> Thanks for the feedback! >>> >>> The reason we extract into a different directory is so that (1) you can >>> have multiple builds and switch between them, and (2) when you >>> install/activate a version, we know if we need to unzip it or not - if the >>> directory exists and is populated, then we have nothing to do. Otherwise >>> we'd always need to unzip the files. But yeah, I do see your point that it >>> makes it harder to just use the build. However, when you do the "activate" >>> step it populates the .emscripten file with the relevant paths - can you >>> get it from there for your build system? >>> >>> The lkgr.json warning should be fixed on emsdk master currently. >>> >>> >>> On Fri, May 24, 2019 at 10:03 AM Floh <[email protected]> wrote: >>> >>>> I'm testing with my cmake-based build setup and I'm having trouble with >>>> the new directory structure of the SDK. >>>> >>>> I need to know the path to the emscripten installation in various >>>> places (among others, a cmake toolchain file), but as far as I can see, >>>> those files are now under a cryptic directory name: >>>> >>>> fastcomp/3b8cff670e9233a6623563add831647e8689a86b/emscripten/emcc >>>> >>>> Before it was something predictable like >>>> >>>> emscripten/incoming/emcc >>>> >>>> I don't want to "pollute" the global environment with environment or >>>> path variables (emcc is not in the path for instance), since I want to >>>> have >>>> different SDK versions side by side. >>>> >>>> Is there any chance we can have a standardized or "predictable" >>>> directory structure? Maybe a link to the actual SDK directory with a >>>> common >>>> name? >>>> >>>> Currently that's a bit of a show stopper for further testing since I >>>> would need to rethink my the entire emscripten integration into my build >>>> system. >>>> >>>> I'm also getting some errors during installation: >>>> >>>> Error downloading URL ' >>>> https://storage.googleapis.com/wasm-llvm/builds/osx/lkgr.json': HTTP >>>> Error 404: Not Found >>>> Error parsing lkgr.json! >>>> [Errno 2] No such file or directory: >>>> '/Users/floh/projects/fips-sdks/osx/emsdk-portable/upstream/lkgr.json' >>>> >>>> Cheers, >>>> -Floh. >>>> >>>> On Friday, 24 May 2019 02:45:05 UTC+2, Alon Zakai wrote: >>>>> >>>>> We have been working hard to replace the current build infrastructure >>>>> with a new system. The new one will be faster, have better support for >>>>> testing, and support the LLVM wasm backend (in fact this change is the >>>>> last >>>>> blocker for us switching to that by default). As of now you can start to >>>>> test this out, if you do the following: >>>>> >>>>> * get latest emsdk master >>>>> * ./emsdk install latest-releases-fastcomp >>>>> * ./emsdk activate latest-releases-fastcomp >>>>> * optionally add the directory to the path with: source ./emsdk_env.sh >>>>> >>>>> That will install a new release tag, 1.38.33, from the new builders. >>>>> >>>>> Note that there is no change to "latest" (or other modes) - this just >>>>> adds this new "latest-releases-fastcomp". But after we are sure things >>>>> work >>>>> well, the goal is to make "latest-releases-fastcomp" replace "latest". >>>>> >>>>> Please let us know if you see any issues with the emsdk or with using >>>>> emcc from that install! >>>>> >>>>> Notes: >>>>> >>>>> * You can replace "fastcomp" with "upstream" to use the LLVM wasm >>>>> backend. We'll have a larger announcement about this soon. >>>>> * The new builders don't download java yet, so closure compiler won't >>>>> work (unless you install it yourself). Aside from that, things should >>>>> work >>>>> just like "latest". >>>>> >>>>> - 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 [email protected]. >>>> To view this discussion on the web visit >>>> https://groups.google.com/d/msgid/emscripten-discuss/fb29262a-bbad-4a29-b818-1d4ad861a5e1%40googlegroups.com >>>> >>>> <https://groups.google.com/d/msgid/emscripten-discuss/fb29262a-bbad-4a29-b818-1d4ad861a5e1%40googlegroups.com?utm_medium=email&utm_source=footer> >>>> . >>>> 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 [email protected] <javascript:>. >> To view this discussion on the web visit >> https://groups.google.com/d/msgid/emscripten-discuss/74cb536f-e540-4a18-ab77-53454ba73292%40googlegroups.com >> >> <https://groups.google.com/d/msgid/emscripten-discuss/74cb536f-e540-4a18-ab77-53454ba73292%40googlegroups.com?utm_medium=email&utm_source=footer> >> . >> 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 [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/emscripten-discuss/29e1de70-011e-4bc0-814d-be293d332142%40googlegroups.com. For more options, visit https://groups.google.com/d/optout.
