Conversation
`nightly-2025-09-14`
RyanGlScott
left a comment
There was a problem hiding this comment.
I suspect (and hopefully CI will confirm) that you'll need to regenerate the vendored-in MIR JSON files in intTests and saw-python/tests/saw/test-files by running make regen-mir-blobs.
|
Or... not? Somehow, this |
|
Oh, hah. This is because of an intended consequence of the changes in GaloisInc/crucible#1619. Previously, SAW would always reject any MIR JSON file whose schema version number doesn't match the expected version. Now, it only checks for version mismatches if the JSON file fails to parse. While the schema did change in the |
|
Ugh. That's convenient here but it's quite easy to imagine a scenario with an old file that parses successfully but incorrectly... |
It is possible for a MIR JSON file to parse even if the schema version doesn't match what is expected, as noticed in GaloisInc/saw-script#2852 (comment). This ensures that the schema version is checked regardless of whether parsing succeeds or not.
|
I've opened GaloisInc/crucible#1661 to fix this particular buglet. |
This should have been regenerated to use `mir-json` schema version 6 as part of the changes in #2852, but this was accidentally left out due to the `test1980` not containing a Makefile. This commit adds such a Makefile and regenerates the MIR JSON.
Brings in the changes from GaloisInc/mir-json#200 and GaloisInc/crucible#1654.