Stream: git-wasmtime

Topic: wasmtime / Issue #1427 there are now two WASI submodules

Wasmtime GitHub notifications bot (Mar 28 2020 at 00:33):

pchickey opened Issue #1427:

As of 6c0c8a46 (#1416), there are two WASI submodules, one at crates/wasi-common/WASI and one at crates/wasi-common/wig/WASI. The commit message doesn't make it clear which of these is intended. I suspect this is something that got messed up in the squash merge?

Wasmtime GitHub notifications bot (Mar 28 2020 at 00:33):

pchickey assigned Issue #1427:

As of 6c0c8a46 (#1416), there are two WASI submodules, one at crates/wasi-common/WASI and one at crates/wasi-common/wig/WASI. The commit message doesn't make it clear which of these is intended. I suspect this is something that got messed up in the squash merge?

Wasmtime GitHub notifications bot (Mar 28 2020 at 00:34):

pchickey edited Issue #1427 (assigned to sunfishcode):

As of 6c0c9a46 (#1416), there are two WASI submodules, one at crates/wasi-common/WASI and one at crates/wasi-common/wig/WASI. The commit message doesn't make it clear which of these is intended. I suspect this is something that got messed up in the squash merge?

Wasmtime GitHub notifications bot (Mar 28 2020 at 00:34):

pchickey edited Issue #1427 (assigned to sunfishcode):

As of 6c0c9a46 (#1416), there are two WASI submodules, one at crates/wasi-common/WASI and one at crates/wasi-common/wig/WASI. The commit message doesn't make it clear which of these is intended. I suspect this is something that got messed up in the squash merge?

Wasmtime GitHub notifications bot (Mar 28 2020 at 00:39):

sunfishcode commented on Issue #1427:

Sorry for the confusion here. The problem is that cargo publish doesn't include wig/WASI in the wasi-common package, causing it to break when built out of tree. An include = directive in Cargo.toml didn't fix it. Adding a WASI symlink fixed it for Mac and Linux but didn't work on Windows.

Since both wasi-common and wig need a WASI directory, for now, the simplest way to fix it was to give them each their own.

Wasmtime GitHub notifications bot (Mar 28 2020 at 01:05):

pchickey commented on Issue #1427:

Dang. Well, wig is hopefully on its last legs, I expect we'll be able to get rid of it within a month or so. In the meantime, we point to the witx path underneath wig. If the one under wasi-common is expected to eventually be the only remaining, should we point there instead?

And, can we add some sort of CI check that both submodules have the same rev checked out, so that they dont accidentally get out-of-sync?

I'll take both of these tasks.

Wasmtime GitHub notifications bot (Mar 30 2020 at 16:07):

alexcrichton commented on Issue #1427:

I managed to cook up https://github.com/bytecodealliance/wasmtime/pull/1434 which I think removes the need for two submodules through "weird cargo hacks"

Wasmtime GitHub notifications bot (Mar 30 2020 at 19:45):

alexcrichton closed Issue #1427 (assigned to sunfishcode):

As of 6c0c9a46 (#1416), there are two WASI submodules, one at crates/wasi-common/WASI and one at crates/wasi-common/wig/WASI. The commit message doesn't make it clear which of these is intended. I suspect this is something that got messed up in the squash merge?

Wasmtime GitHub notifications bot (Mar 30 2020 at 20:42):

pchickey commented on Issue #1427:

Thank you!

Last updated: Oct 23 2024 at 20:03 UTC