I added a simple test of embind with groups in https://github.com/kripken/emscripten/commit/8795cbd3adb5da740a15cd4971f6a61054642763 , and I don't see the problem yet. Do you have a testcase?
On Mon, Feb 22, 2016 at 5:32 PM, arnab choudhury <[email protected]> wrote: > Thanks guys - there seems to be an issue with the --start-group > --end-group feature when used with the --bind option. Emcc tries to use the > same flags passed in to try and compile bind.cpp and fails because > clang++.exe doesn't recognize the --start-group and --end-group flags. > > > On Thursday, February 18, 2016 at 2:27:20 AM UTC-8, Floh wrote: >> >> This is common problem on GCC too. I got rid of the problem by defining a >> proper dependency tree for all my code modules for cmake. E.g. in cmake you >> can define dependencies between libraries (e.g. lib B depends on lib A, and >> then when linking the executables you only need to tell cmake to link with >> B and A will automatically be linked because B depends on it. >> >> Setting up all those inter-module dependencies will also automagically >> fix the link-order problem. >> >> -Floh. >> >> Am Donnerstag, 18. Februar 2016 01:56:22 UTC+1 schrieb Alon Zakai: >>> >>> Yes, this is how .a linking works, it's pretty confusing actually. It >>> selects .o files out of the the .a based on what is used. So the order >>> definitely can matter. To make it not matter, there are the --start-group >>> etc. flags, which keep looping until everything is resolved. >>> >>> .a files can make linking faster since they avoid .o files in a quick >>> way. But it's usually fine to avoid .a files and just use .bc/.o files that >>> just contain plain bitcode. >>> >>> On Wed, Feb 17, 2016 at 4:30 PM, arnab choudhury <[email protected]> >>> wrote: >>> >>>> Hey guys >>>> >>>> I just ran into an issue where I was seeing unresolved externals when >>>> using Emscripten to transpile my test c++ file to JS. After digging for a >>>> while, I figured out that the order of includes of the .a files is >>>> important. For example, if you have 2 static libraries foo.a and bar.a, and >>>> bar.a has a dependency on foo.a, you must necessarily specify bar.a before >>>> foo.a when passing in command line arguments to emcc. >>>> >>>> In the example above, >>>> emcc foo.a bar.a test.o -o test.js will produce unresolved externals >>>> while >>>> emcc bar.a foo.a test.o -o test.js will not. >>>> >>>> Is this a known issue? I can provide more specifics if needed. >>>> >>>> Thanks! >>>> -Arnab >>>> >>>> -- >>>> 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. >>>> >>> >>> -- > 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. > -- 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.
