Add wasm-3.0 branch to the sync'd tests here #110
Conversation
Update a few variables to handle that it's a branch of the `spec` repo itself.
spec:
WebAssembly/spec@bd2aa85d
memory64:
WebAssembly/memory64@334f93f9
custom-page-sizes:
WebAssembly/custom-page-sizes@ba05ecdd
wasm-3.0:
WebAssembly/spec@90cbd509
This change was automatically generated by `update-testsuite.sh`
Some minor updates in preparation for WebAssembly/testsuite#110: * Don't limit the size of tables in the validator and instead defer such maximal validation to runtimes themselves. * Parse the `(pagesize N)` syntax on "inline memories" * Always parse table/memory limits as 64-bit integers * Stop matching spec test suite error messages This last point is something I've tried to hold off on doing for a long time but as the giant `error_matches` function continues to grow and become more unwieldy it has become less and less tenable to do this. Overall it doesn't seem too beneficial to keep maintaining this especially in the face of numerous proposals, so instead basically remove it entirely.
update-testsuite.sh
Outdated
| @@ -80,6 +85,8 @@ merge_with_spec() { | |||
|
|
|||
| set_upstream ${repo} | |||
|
|
|||
| [ "${repo}" == "wasm-3.0" ] && return | |||
There was a problem hiding this comment.
Can you add a comment here describing why we early return in this case?
There was a problem hiding this comment.
In commenting this I realized it was actually easier to just go ahead and handle this, so it no longer early-returns
* Prepare to update spec test suite Some minor updates in preparation for WebAssembly/testsuite#110: * Don't limit the size of tables in the validator and instead defer such maximal validation to runtimes themselves. * Parse the `(pagesize N)` syntax on "inline memories" * Always parse table/memory limits as 64-bit integers * Stop matching spec test suite error messages This last point is something I've tried to hold off on doing for a long time but as the giant `error_matches` function continues to grow and become more unwieldy it has become less and less tenable to do this. Overall it doesn't seem too beneficial to keep maintaining this especially in the face of numerous proposals, so instead basically remove it entirely. * Rename test
|
I think they're pseudo-folded in here perhaps? For example |
|
I felt like giving it a spin so I've pushed a rewrite from bash to python in this PR as well. I'm happy to split that out to a separate PR if preferred. I've tested it locally by ensuring that if I run it it does nothing (e.g. no updates) and then if I revert the updates made in this PR it generates the same set of updates. One major change from before is that it no longer runs |
Oh wow! Nice work. I do think it might be nice to land this functional change first, before reviewing the re-write. Can we land the version prior to the re-write? |
alexcrichton
force-pushed
the
wasm3-suite
branch
from
128761d to
765e40c
Compare
|
Sounds good, rolled back that commit and I'll post it after this merges |
|
Oh I'll interpret @sbc100 your approval as "ok for me to merge" so I'm going to merge. (sorry I forget which projects prefer others merge vs which are ok with self-merging) |
I've tried to hack this into update-testsuite.sh but I probably made a mistake here or there. I believe this handles the removed tests in #109 though.