Sam Clegg

> is a real mmap implementation Obviously you can't make a "real" mmap implementation without virtual memory right? One would still basically need to fake mmap on top of `sbkr`...

I imagine what is happening is that those lines around line 10754 are not very well tested and are not being compiled on the native compile because `CATCH_CONFIG_POSIX_SIGNALS` is being...

To be clear, this issue also occurs if emscripten and the project are on the same drive right? You mentioned on the mailing list that the source map contained a...

What does that `emscripten_get_heap_size.c` line looks like in the broken source map?

That looks correct to me. Do you know where the bad path (`myArea/myApp/C:/Users/myName/emsdk/upstream/emscripten`) path in your original report comes from?

I'm not sure what you mean by "localhost" here. Are you referring to the http server you are using to serve your source files? If its asked to serve `C:/Users/myName/emsdk/upstream/emscripten`...

> so it uses the local file path, which can't work. Why do you say it cannot work? Its seems like a limitation of the server "Internet Information Server"? Perhaps...

> Thanks for the suggestion, I will have a look at that. Chrome turns the absolute path in the map into `file:myArea/myApp/C:/Users/diete/emsdk/upstream/emscripten` Oh I see.. that does seem wrong. So...

Hmm, it seems that having a unified way to specify exports is somewhat orthogonal to whether the names are mangled. But I can see that if you are going to...

I've been thinking about a plan for how to make this happen: 1. Add a new setting `MANGLE_NATIVE_SYMBOLS`, which defaults to `true` (perhaps not in `-sSTRICT` mode). 2. Generate unmangled...