Increase testability of constraint failures - #351
Open
jan-ferdinand wants to merge 15 commits into
Open
jan-ferdinand wants to merge 15 commits into
jan-ferdinand wants to merge 15 commits into
Conversation
jan-ferdinand
force-pushed
the
master
branch
3 times, most recently
from
May 12, 2025 05:38
30f17a2 to
9407c96
Compare
jan-ferdinand
force-pushed
the
constraint_tests
branch
2 times, most recently
from
June 5, 2025 14:08
9250c93 to
0f4d144
Compare
jan-ferdinand
force-pushed
the
constraint_tests
branch
from
February 13, 2026 07:42
0f4d144 to
89af390
Compare
Both the “Fast Reed-Solomon IOPP” (“FRI”) and “Shift to Improve Rate” (“STIR”) are proof systems that establish the low-degreeness of a polynomial. They each have their own advantages and disadvantages. In particular, for our STARK and when the execution trace is short, the size of the proof is significantly larger when using STIR instead of FRI. For longer execution traces, this is reversed. In addition, STIR proofs allow for some exciting future enhancements to the overall proof system. This commit also moves away from the disproven conjectured bound that was in use previously: Some recent results have proven that Reed-Solomon codes do not exhibit a proximity gap at the capacity-bound. The list size of Reed-Solomon codes with proximity parameter beyond the Johnson bound is a famous open problem; the security of any bound beyond the Johnson bound is conjectured at best for the time being. BREAKING CHANGE: In order to add STIR as an LDT, the version of the proof system is bumped from 2 to 3. This means that old proofs are considered invalid by this version of Triton VM.
While there is a clippy lint for this combination, I find `map` + `unwrap_or` more readable than `map_or`. changelog: ignore
changelog: ignore
changelog: ignore
(Where needed.) changelog: ignore
If Triton VM is configured to cache as much as possible but cannot allocate the required amount of memory, return an error instead of panicking. Note that there is no guarantee that this behavior will trigger: some operating systems do “memory overcommitment”, seemingly providing access to memory that is not, in fact, available. In such cases, the out-of-memory error will not be triggered and the operating system instead terminates the process consuming the most RAM as soon as the machine's RAM is exhausted.
changelog: ignore
A future update to `rustc` will turn this into a hard error. changelog: ignore
Also, reduce overall test times by limiting the number of cases of some long-running proptests. changelog: ignore
Part of this change is to run the link checker weekly, not daily. Due to some recent changes somewhere on the internet (I really don't know where), the link checking workflow has become quite a bit more noisy than it used to be, creating issues that resolve themselves within a day or two. With the new behavior, if there are no links known to be broken, then the link checker runs only once a week. Otherwise, if there are broken links in the repository somewhere, the link checker runs once a day. The intention of this change is a reduction of the overall noise. The weekly link checker run might create a noisy issue, but at most one per week. The daily runs keep that issue up to date, accurately informing in case of a real problem. Also, exclude eprint.iacr.org and iacr.org/crypoDB from link checking: Due to heavily restricted automated access to the eprint archive, forgo checking whether the pointed-to pages are still available. While it is sad to diminish the effectiveness of the automated link checker, I do understand their reasons for limiting automated access. changelog: ignore
jan-ferdinand
force-pushed
the
constraint_tests
branch
from
May 11, 2026 12:27
89af390 to
eddd2a2
Compare
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
On unfulfilled constraints, don't just print strings to stderr, but return a dedicated type.
Also, restructure test helper functions. In particular, return fewer tuples, use dedicated types instead.