On Mon, Aug 12, 2019 at 4:10 PM Piotr Paczkowski <[email protected]> wrote:

> Hi,
> I'm a maintainer of https://hub.docker.com/r/trzeci/emscripten/ and I
> need some hand with last changes for Emscripten backend
>
>
Great, getting wasm backend support there is very important I think!


> My current observation is that 1.38.31 was the last SDK version which was
> possible to compile fastcomp from sources, then every other version
> includes this 'frozen' version. Than is if I install 1.38.40 (non-upstream)
> I'm getting binaries of clang that I got with 1.38.31. Is that correct?
>

Yes, fastcomp hasn't changed since 1.38.31 - it's the same code since then
(we would fix any urgent things if necessary, but there haven't been any,
and otherwise our focus is on the new backend).


> In case of the upstream version I'm also getting aside fastcomp version:
> ```
> ./upstream/fastcomp/bin/clang --version
> clang version 6.0.1
> (/b/s/w/ir/cache/git/chromium.googlesource.com-external-github.com-emscripten--core-emscripten--fastcomp--clang
> 98df4be387dde3e3918fa5bbb5fc43e1a0e1daac)
> (/b/s/w/ir/cache/git/chromium.googlesource.com-external-github.com-emscripten--core-emscripten--fastcomp
> 1b4148f39a69c7fc62edadd85e4122b68694dfb7) (emscripten 1.38.31 : 1.38.31)
> ```
> Why is that needed still?
>
>
I'm not sure what you mean by "aside" in that sentence? And not sure I
understand the question, sorry.

The directory structure is that fastcomp binaries are in a subdirectory, so
./upstream/fastcomp/bin/clang is fastcomp, while ./upstream/bin/clang
(without /fastcomp/) will be upstream. So that --version output shows
fastcomp (which uses old 6.0.1 clang, unlike upstream which is 10.0.0).

Last question about the upstream clang version which is ATM: clang version
> 10.0.0
> (/b/s/w/ir/cache/git/chromium.googlesource.com-external-github.com-llvm-llvm--project
> ea134f221f2a5c075b7539876a444b4a07362912).
> - is there something special with served binaries (like special
> compilation flags, or just a regular clang 10 release)? Is it needed to be
> on clang 10.0.0?
>

A special revision of clang/llvm is needed. For example a change to wasm-ld
in LLVM will require a corresponding change in emscripten's code that
drives wasm-ld. See the DEPS notes and links here:

https://github.com/emscripten-core/emscripten/blob/incoming/docs/process.md#packaging-emscripten

Basically, that emscripten-releases repo has all the info for which
revision should work with which other revision. (See also notes on which
llvm tools are needed, like wasm-ld etc.)

Btw, does the Docker image use the emsdk? In that case these details
wouldn't be necessary. It will also generate a .emscripten file looking at
the right bin/ directory etc. I wonder if the Docker image could just do
that?

What I notice is that regardless if I install 1.38.34-upstream or
> 1.38.41-upstream - I'm getting exactly the same binaries of clang
>
>
The fastcomp binaries will be the same, but the binaries in the bin/
(without fastcomp/) should not be.

If that doesn't answer your question, what emsdk (or other) commands
specifically are you doing?

- Alon

Any help with solving/confirming issues from above will be helpful to get
> back on track with docker images.
>
>
> Cheers!
>
> --
> 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/e8741ec1-f53e-4a39-afc1-5461a8cfee29%40googlegroups.com
> <https://groups.google.com/d/msgid/emscripten-discuss/e8741ec1-f53e-4a39-afc1-5461a8cfee29%40googlegroups.com?utm_medium=email&utm_source=footer>
> .
>

-- 
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/CAEX4NpSShXPsPqnYfiyPDNXKGHCxMvAi2BqiZJzKzXpEjBqDWg%40mail.gmail.com.

Reply via email to