abrown commented on issue #7433:
Drive-by: I'd agree with the documentation piece @elliottt mentioned and would go a bit further to note that this environment variable approach (which I also used for the
differential
target) means the corpus is invalid if it was generated in a different environment — I have forgotten this fact before (e.g., when sending in a reproducer).
afonso360 commented on issue #7433:
:+1: Added some docs in the README.md file.
note that this environment variable approach (which I also used for the differential target) means the corpus is invalid if it was generated in a different environment
IIRC @elliottt made it so that having different instructions available wouldn't automatically invalidate the whole corpus (i.e. previous inputs still work). We used to have this issue a lot when first adding instructions, where each PR would invalidate the entire OSS-Fuzz corpus.
github-actions[bot] commented on issue #7433:
Subscribe to Label Action
cc @fitzgen
<details>
This issue or pull request has been labeled: "cranelift", "fuzzing"Thus the following users have been cc'd because of the following labels:
- fitzgen: fuzzing
To subscribe or unsubscribe from this label, edit the <code>.github/subscribe-to-label.json</code> configuration file.
Learn more.
</details>
afonso360 edited a comment on issue #7433:
:+1: Added some docs in the README.md file.
note that this environment variable approach (which I also used for the differential target) means the corpus is invalid if it was generated in a different environment
IIRC @elliottt made it so that having different instructions available wouldn't automatically invalidate the whole corpus (i.e. previous inputs still work). We used to have this issue a lot when first adding instructions, where each PR would invalidate the entire OSS-Fuzz corpus.
Edit: It looks like we do that, but only at a later stage for individual arches. So this does invalidate the corpus.
Last updated: Dec 23 2024 at 12:05 UTC