rs545837

I don't think it's only C++ though, Python decorators, Rust macros, almost every language's compiler write code you never see. C++ just does more of it and makes it harder to guess. Being able to look at the expanded version is useful in any language that does this.

wow this is fascinating read

Damn this was really fun to use.

I agree 100% with this thinking approach, I've been working in this domain for quite a few months now.

The right granularity for agents isn't files or lines, it's entities: functions, classes, methods. That's how both humans and agents actually think about code.

We built sem(Ataraxy-Labs/sem) which extracts entities from 30+ languages via tree-sitter and builds a cross-file dependency graph, so building semantic version control and semantic diff. weave (same org) takes it further and does git merges at entity level. Matches functions by name, merges their bodies independently.

The dependency graph also answers questions LLMs can't. I love the analysis based on ASTs.

yeah definitely plastic scm has always been my inspiration, just trying to revive it.

Oh this is cool. The Yjs as storage backend trick is clever, you basically get CRDT sync for free without having to build your own replication layer. And the pluggable storage means you can develop against in-memory and then flip to YGraph for collab mode without touching your queries. That's a nice developer experience.

The live queries also caught my eye. Having traversals auto reexecute when data changes sounds straightforward until you realize the underlying data is being merged from multiple peers concurrently. Getting that right without stale reads or phantom edges is genuinely hard.

I've been researching on something like this in a similar space but for source code, therefore built a tool called Weave(https://github.com/Ataraxy-Labs/weave) for entity level merges for git. Instead of merging lines of text, it extracts functions, classes, and methods, builds a dependency graph between them, and merges at that level.

Seeing codemix makes me think there might be something interesting here. Right now our entity graph and our CRDT state are two separate things. The graph lives our analysis engine and the CRDT lives in different crate. If something like @codemix/graph could unify those, you'd have a single data structure where the entity dependency graph is the CRDT.

Think two agents working on the same codebase at the same time. Agent A is refactoring the auth module, Agent B is adding a new API endpoint that imports from auth. Separate worktrees, separate branches, but they're touching overlapping code.

ingle agent per feature works great today. But as agents get faster and cheaper, the bottleneck shifts to, how many agents can work on one repo simultaneously without stepping on each other.

You could agree that the PR is the meaningful unit for shipping, but push back gently that for agents working in parallel, the commit/changeset level matters more than it used to because agents don't coordinate the way humans do. Multiple agents touching the same repo need finer-grained units of change than "the whole PR."

jj is genuinely great and I think it deserves way more adoption than it has right now. The mental model is so much cleaner than git, undo actually works the way you'd expect it to, and working with stacked changes feels natural instead of that constant low-grade anxiety of actually breaking something. It's probably the best frontend for version control that exists today.

For the last few months though I've been thinking a lot about what you said at the end there. What if version control actually understood the code it was tracking, not as lines of text but as the actual structures we write and think in, functions, classes, methods, the real building blocks? A rename happening on one branch and an unrelated function addition on another aren't a real conflict in any meaningful sense, they only look like one because every tool we have today treats source code as flat text files.

For enhancing this kind of structural intelligence I started working on https://github.com/ataraxy-labs/sem, which uses tree-sitter to parse code into semantic entities and operates at that level instead of lines. When you start thinking of code not as text there's another dimension where things can go, even a lot of logic at the comiler level with call graphs becomes useful.

This is awesome honestly, Stacked PRs are one of those features that feels obvious in hindsight. Breaking a n-line PR into 3 focused layers where each one is independently reviewable is a huge win for both the author and reviewer. The native GitHub UI with the stack navigator is the right call too, and there's no reason this should require a third-party tool.

One thing I keep thinking about in this same direction: even within a single layer of a stack, line-level diffs are still noisy. You rename a function and update x call sites, the diff shows y changed lines. A reviewer has to mentally reconstruct "oh this is just a rename" from raw red/green text.

Semantic diffing (showing which functions, classes, methods were added/modified/deleted/moved) would pair really well with stacks. Each layer of the stack becomes even easier to review when the diff tells you "modified function X, added function Y" instead of just showing changed lines.

I've been researching something in this direction, https://ataraxy-labs.github.io/sem/. It does entity-level diffs, blame, and impact analysis. Would love to see forges like GitHub move in this direction natively. Stacked PRs solve the too much at once problem. Semantic diffs solve the "what actually changed" problem. Together they'd make code review dramatically better.

One cheap optimization for the compile overhead case: skip commits that only touch files unrelated to the failing test. If you know the test's dependency chain, any commit that doesn't touch that chain gets prior weight zero. Equivalent to git bisect skip but automatic. Cuts the search space before you compile anything.

This is a real pain point. One thing that helps: when an LLM agent makes changes across multiple commits, look at what it actually touched structurally. Often the agent adds a feature in commit 5 but subtly breaks something in commit 3 by changing a shared function it didn't fully understand.

You're right, at 300 tests bayesect converges to ~97-100% across the board. I reran with calibration.py and confirmed.

Went a step further and tested graph-weighted priors (per-commit weight proportional to transitive dependents, Pareto-distributed). The prior helps in the budget-constrained regime:

128 commits, 500 trials:

Budget=50, 70/30: uniform 22% → graph 33% Budget=50, 80/20: uniform 71% → graph 77% Budget=100, 70/30: uniform 56% → graph 65% At 300 tests the gap disappears since there's enough data to converge anyway. The prior is worth a few bits, which matters when bits are scarce.

Script: https://gist.github.com/rs545837/b3266ecf22e12726f0d55c56466...

Yes, bayesect accuracy increases with more iterations. The comparison was at a fixed budget(300 test runs) when I was running. Sorry should have clarified more on that.

Really fun work, and the writeup on the math is great. The Beta-Bernoulli conjugacy trick making the marginal likelihood closed-form is elegant.

We ran benchmarks comparing bisect vs bayesect across flakiness levels. At 90/10, bisect drops to ~44% accuracy while bayesect holds at ~96%. At 70/30 it's 9% vs 67%. The entropy-minimization selection is key here since naive median splitting converges much slower.

One thing we found, you can squeeze out another 10-15% accuracy by weighting the prior with code structure. Commits that change highly-connected functions (many transitive dependents in the call graph) are more likely culprits than commits touching isolated code. That prior is free, zero test runs needed.

Information-theoretically, the structural prior gives you I_prior bits before running any test, reducing the total tests needed from log2(n)/D_KL to (log2(n) - I_prior)/D_KL. On 1024-commit repos with 80/20 flakiness: 92% accuracy with graph priors vs 85% pure bayesect vs 10% git bisect.

We're building this into sem (https://github.com/ataraxy-labs/sem), which has an entity dependency graph that provides the structural signal.

usually the whole discussion has been around line-level vs commit-level history, but there's a layer nobody's talking about, and I have been exploring it here these days with https://github.com/Ataraxy-Labs/sem. It gives you entity-level version control. It parses your code into functions, classes, methods using tree-sitter (12 languages so far), computes a structural hash for each entity, and builds a cross-file dependency graph. So sem diff HEAD~1 doesn't give you "+3 -2 in tax.py", it gives you "calculate_tax signature changed, 47 dependents, 3 callers will break". The key insight is distinguishing signature changes from body changes.

You can, but it's slow, expensive, and hallucinates. An LLM looking at a raw diff might miss a renamed function or invent a dependency that doesn't exist. sem does it structurally: parses both sides with tree-sitter, computes structural hashes, walks the real dependency graph. If you want to layer an LLM on top for summarization, you're feeding it 10 entities instead of 500 lines of unified diff.

Right, sem gives you both. sem diff --verbose shows the full before/after body of each changed entity. The entity-level view tells you what changed and what's affected. The line-level detail is still there when you need it.

We've been building an open source tool called sem (https://github.com/ataraxy-labs/sem) that takes this one level further: entity-level diffs instead of AST-level.

Instead of showing you which syntax nodes changed, it shows you which functions, classes, and methods changed, classifies the change (text-only, syntax, functional), and walks a dependency graph to tell you the blast radius.

The delta + difftastic integration problem in that issue is interesting because sem already has the pieces both sides need, before/after content with full context for every changed entity, plus structured JSON output. The blocker in #535 is that difftastic's JSON doesn't include surrounding context. sem's output includes complete entity bodies by default.

Would love to collaborate on a common interchange format if anyone from the delta or difftastic projects is interested. Entity-level granularity sits naturally above AST-level diffs and below file-level diffs, and having a standard way to represent "what changed and what depends on it" would be useful for the whole ecosystem.

That's a really sharp parallel. "Did behavior change" is exactly the question in both cases, and the surface-level representation lies to you in both. We normalize ASTs before hashing so reformatting or renaming a local variable doesn't register as a change. Curious what normalization looks like on the agent observability side, feels like a harder problem when the output is natural language instead of code.

Great, did you opensource it?

Pretty much, yeah. sem graph builds a cross-file entity dependency graph and sem impact does transitive traversal on it. So you can ask "if I change this function, what breaks?" and get a deterministic answer without sending anything to an LLM.

difftastic is solid. The difference is roughly: syntax-aware (difftastic) knows what changed in the tree, sem knows which entity changed and whether it actually matters. difftastic will show you that a node in the AST moved. sem will tell you "the function processOrder was modified, and 3 other functions across 2 files depend on it." difftastic is a better diff. sem is trying to be a different layer on top of git entirely.

Ah right, array ordering not key ordering. That's a different beast. You're making a deliberate semantic choice because you know your consumers are order-invariant. We can't really do that at our level since function ordering in code is usually meaningful to the language. Your use case needs domain knowledge about the consumer, which is exactly why an option makes sense there.

If you do end up comparing semantic toolings I'd love to hear what you find. The space is weirdly fragmented between syntax-aware, normalized-AST, and domain-specific (dyff/jd). Everyone calls it "semantic" but they're solving pretty different problems.

Thanks for the request, Yeah I can work on this.

Exactly. The only reason line-level diffs survived this long is that text is the lowest common denominator. Once you have fast enough parsers (tree-sitter parses most files in under 1ms), there's no reason to stay at the line level.

sem already does this. sem graph builds a cross-file entity dependency graph and sem impact tells you "if this function changes, these other entities across these many files are affected." It's transitive too, follows the full call chain.

You're right that "semantic" is doing some heavy lifting in the name. We use it to mean "understands code structure" rather than "understands code meaning." sem knows that something like "validateToken" is a function and tracks it as an entity across versions, but it doesn't know what validation means. For the merge use case (weave - https://ataraxy-labs.github.io/weave/), that level of understanding is enough to resolve 100% of false conflicts on our benchmarks. LLM-powered semantic understanding is the next layer, and that's what our review tool (inspect) does, it uses sem's entity extraction to triage what an LLM should look at.

Different project, same author. sem has been sem since the first commit. Beagle looks interesting, storing ASTs directly in a KV store is a different approach. sem stays on top of git so there's zero migration cost, you keep your existing repo and workflows.

Fair point on AST vs semantic. sem sits somewhere in between. It doesn't go as far as checking compiled output equivalence, but it does normalize the AST before hashing (we call it structural_hash), so purely cosmetic changes like reformatting or renaming a local variable won't show as a diff. The goal isn't "would the compiler produce the same binary" but "did the developer change the behavior of this entity." For most practical cases that's the useful boundary. The YAML/JSON ordering point is interesting, we handle JSON keys as entities so reordering doesn't conflict during merges.

By the way creator here.