FluidFramework — Engineering Performance
32 engineers all time · Jan 2025 – Nov 2026 · built 2026-11-22 · GitHub
Performance snapshot
Today's rolling 90-day reading for FluidFramework, compared with the start of the series. Pick a window to move that comparison point.
Avg. perf / dev / mo
+127.4%
0.77 → 1.76 ETV
Active engineers
−20.0%
30.0 → 24.0
Features
+9.7pp
12.9% → 22.6%
vs. Microsoft
0.16x
0.57x → 0.16x · −84% below
FluidFramework vs. Microsoft
Per-engineer ETV for FluidFramework against Microsoft as a whole. Both lines are 90-day rolling averages scaled to a 30-day month, so they share one axis and can be read against each other at any point. Pick a window to zoom the chart to it.
Performance Composition
Each month's output split by type of work: Features (new value), Maintenance (sustaining systems), Tests, Docs, and Fixes (rework). The yellow line is output per engineer, so when it rises each engineer is delivering more, whatever the team size did. Unit: Engineering Throughput Value (ETV).
Engineering capacity
Effective engineers behind FluidFramework, against its pre-AI baseline. Each subject has its own: FluidFramework's is 0.77 ETV / dev / mo, its first reading in Q1 2025. Per-engineer ETV divided by that gives a capacity multiple, and that multiple applied to the engineers active in the trailing 90 days turns it into engineer-equivalents. The line is the real headcount, so the gap between line and area is what the leverage is worth. Because each baseline is its own, every subject opens at 1.0x on its first day: multiples measure improvement and are not comparable between subjects.
Knowledge concentration
How dependent is this repo on a small number of engineers? Higher top-1 share = higher key-person risk.
Craig Macomber (Microsoft) owns 19.4 % of commits.
Behind the numbers
Written summary of the work completed each month.
No monthly reports available yet.
Top engineers
Most impactful commits
Top 10 by ETV in the all-time window.
- 12.3ETVtree: Refactor detached changeset model (mark 2) (#28667) ## Description This PR introduces a new editing model at the rebaser layer (higher/public layers will be updated to leverage this new model in a separate PR). The purpose of this new editing model is to support the following new kinds of edits: * Editing a detached tree (without attaching it first) * Attaching a detached tree (under some conditions) This is in contrast to the current editing model where, once a tree is known to be removed (i.e., detached), it cannot be reinserted (i.e., reattached) except by reverting the commit that removed/detached it (which only the issuer of that commit can do) and it cannot be edited until it is reinserted. The new editing model aims to greatly simplify some editing scenarios. For example, moving a node from A to B currently requires a one-step move operation that humans and LLMs alike tend to get wrong. The new editing model will make it possible to move the node by first detaching it from location A and re-attaching it at location B. ## Changes to Persisted Formats New persisted formats have been introduced at multiple layers, but this PR does not yet enable them for writing in production environments. This PR does not therefore make new format commitments. The new formats can be used for testing through `SharedTreeOptions.enableDetachedRootEditing`. ## Breaking Changes This PR is not intended to have breaking changes but is worthy of scrutiny since it changes how existing formats are encoded and decoded. ## Authors @alex-pardes @yann-achard-MS @jason-ha --------- Co-authored-by: alex-pardes <105307164+alex-pardes@users.noreply.github.com> Co-authored-by: Jason Hartman <jasonha@microsoft.com>yann-achard-MS · 34ab1001 · 2026-11-13
- 6.9ETVrefactor(markdown-magic): Replatform onto remark/mdast (#28098) This pattern unlocks a lot more flexibility in how we do generation / content copying. New functionality included in this PR: - Support for `.mdx` files (like we use in the website) - Automatic heading level detection when creating contents that include headings. Future improvements that will be possible: - Automatic file path link re-mapping (so you can safely embed contents that include links). Also updates the code to use TypeScript.Joshua Smithrud · 44e6a4cd · 2026-09-09
- 3.6ETVdocs: Update Client Requirements template for 3.0 release (#27802) Updates the Client Requirements template with updates for 3.0. This content is embedded in the majority of our library package READMEs. [AB#78964](https://dev.azure.com/fluidframework/235294da-091d-4c29-84fc-cdfc3d90890b/_workitems/edit/78964)Joshua Smithrud · a12075a9 · 2026-08-24
- 3.4ETVrefactor(examples): Add explicit function return types (#26074) Part of a multi-PR effort in preparation for enabling global eslint enforcement.Joshua Smithrud · bf85f1d4 · 2025-12-20
- 2.7ETVtest(api-markdown-documenter): Add "deep" config end-to-end test cases (#23586) Adds a test case that leverages new hierarchy options to create a "deep" suite configuration where documents that generate folder hierarchy yield documents named "index" _inside_ their folder,Joshua Smithrud · 5f4097ca · 2025-01-17
- 2.7ETVfeat(tree): Support sharded id-compressor in sandbox guests (#28360) The sandbox Guest now uses a serialized child ID space shard instead of sharing the Host's live ID compressor. The Host keeps its runtime compressor and sends the child shard with the initial tree and retained changes. Guest changes carry child ID progress so the Host can decode new IDs. Host updates carry parent progress, and ordered messages deliver finalized ID ranges before changes that need them. The Host returns the Guest's ID space shard to its ID compressor upon its closure. --------- Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>Joshua Smithrud · b1f94e72 · 2026-10-03
- 2.6ETVperf(tree): share stored schemas across policies without staging (#28530) ## Description Cache view-schema graphs staging reachability to reuse stored schemas across policies when no staging is present. Add independent catalog expectations and regression coverage for recursive schemas and cache population order.Craig Macomber (Microsoft) · 4890b806 · 2026-11-03
- 2.6ETVbuild(client): generate eslint 9 configs for all packages (#25989) This change updates the shared eslint 9 config to be TypeScript and updates the script in scripts/generate-flat-eslint-configs.ts to support a `--typescript` flag to output typescript per-package configs instead of ESM. Also adds a `--finalize` flag that will do some additional cleanup once packages are fully migrated to eslint 9. Finally, there are lots of fixes to the script to handle ignore entries, inheritance from other configs, etc. The script now produces configs that work in my testing without any modifications. Even if that doesn't hold true once we are truly migrating (see #25932), merging this beforehand will simplify the review of those changes. All the configs were generated by the script using this command: ``` tsx scripts/generate-flat-eslint-configs.ts --typescript ```Tyler Butler · cfbf9607 · 2025-12-09
- 2.4ETVfeat(tree): staged allowed types (#25116) ## Description Adds `staged` API to `SchemaFactoryAlpha`. This creates an allowed type in the view schema that may or may not be included int he stored schema. Continuation of https://github.com/microsoft/FluidFramework/pull/24631, with a new PR from my fork so I can have push permissions. Major changes: - Adds SchemaFactoryAlpha.staged - There is no longer a single choice for how to derive a stored schema from a view schema: - Different use cases have been split into different APIs for example (toInitalSchema, toUpgradeSchema) - The underlying toStoredSchema now takes options to control how each staged schema is handled. - Unhydrated content always permits staged content (with the exception of clone): this ensures that export/import round tripping of staged content works. - testDocuments test suite has been expanded so existing round trip tests cover the above. Known issues/limitations: - Recursive types are not supported. Tracked by https://dev.azure.com/fluidframework/internal/_workitems/edit/45711 - Clone should produce nodes with a union of source context and staged types so that both unknown optional fields work, and new staged types can be inserted. Tracked by https://dev.azure.com/fluidframework/internal/_workitems/edit/45725 but also partially hidden by https://dev.azure.com/fluidframework/internal/_workitems/edit/45723 which covers how inserting the out of schema types does not error in unhydrated context currently. This is ok, as we catch them when inserting into hydrated documents and thus the two pugs combine to make the desired behavior, but itn't great, and could cause other issue due to violating internal invariants. Some of the new clone tests cover this. - Some places, mainly those not using alpha typer, can't accept annotated allowed types and thus don't accept staged types. Examples include the root of the tree view config, recursiveObject fields, and likely more. Many of these can be worked around using dedicated alpha APIs and/or wrapping the implicit field schema in an explicit one using SchemaFactoryAlpha.required. Some cases, like recursiveArray do not have a viable workaround, and can be addressed in future work, possibly after stabilizing annotated allowed types.Craig Macomber (Microsoft) · 59baf03a · 2025-08-07
- 2.4ETVfeat(tree): add persisted commit metadata (#28064) ## Description Applications can now attach arbitrary, JSON-serializable metadata to the commit that a transaction produces, replicate it to collaborating clients, and persist it in the document. Write it via the new `customMetadata` field on `RunTransactionParamsAlpha`: ```typescript view.runTransaction( () => { view.root.insertAtEnd("new item"); }, { customMetadata: { author: "alice", intent: "add-item" } }, ); ``` Read it back while walking the branch's [history](https://fluidframework.com/docs/api/fluid-framework/treebranchhistory-interface), via the new `custom` property on `TreeBranchCommitMetadata`: ```typescript for ( let commit = view.branchHistory.getHead(); commit !== undefined; commit = commit.getParent() ) { const metadata = commit.custom; } ``` Because a commit may be produced by nested transactions, each of which may supply metadata, `custom` is the flattened combination of them all (outermost wins on conflicting keys). The structural view is available as `commit.customTree`, a `CustomMetadataTree` mirroring the transaction nesting — the same relationship `labels.tree` has to a change's label set. The metadata lives directly on the commit — on `GraphCommit.customMetadata` in memory and inline on the commits in the `EditManager` summary. That is what makes its lifetime automatically match the commit's: once the commit is trimmed from the trunk, the metadata goes with it, so there is no separate index to populate, reconcile, or prune. Both the op format and the summary format gain a `v7` (`MessageFormatVersion.v7` and `EditManagerFormatVersion.v7`), written only when `minVersionForCollab` is `2.117.0` or later (gated by `FluidClientVersion.v2_117`). That floor is a declaration rather than an enforcement mechanism: a client too old for v7 fails cleanly with an unsupported-version error when it *reaches* v7 data, so adopting this requires deploying v7-capable readers everywhere before raising the floor. The changeset spells out the rollout sequence. A revert may now be performed inside a transaction provided it is that transaction's only change, which lets the revert be given its own metadata. Attempting any other change in such a transaction throws an error. ## Reviewer Guidance The review process is outlined on [this wiki page](https://github.com/microsoft/FluidFramework/wiki/PR-Guidelines#guidelines). ### The `GraphCommit.customMetadata` property is required, not optional This is the load-bearing decision. Two places rebuild a commit from its parts rather than spreading it (`mintCommit` and `rebaseBranch`), and an optional property would let both silently drop the metadata. Declaring it required turns each into a compile error, so the type system enumerates every site that has to make a decision. Where a rebuilt commit is the *same logical commit* as its source, the property is propagated. `undefined` is used only where a genuinely new commit is minted — the inverse commit produced by reverting, rollback commits, the synthetic root/trunk-base commits, and edits from the editor that a transaction has not yet annotated. ### The metadata tree mirrors transaction nesting Each transaction in a nested stack contributes a node to a `CustomMetadataTree`. The root node is the outermost transaction, and nested transactions add child nodes. Aborted nested transactions remove their node again, so they never contribute. The single commit produced by the stack carries the tree, from which the flattened `custom` view is derived. ### Compatibility The `v7` codecs are only selected when `minVersionForCollab >= 2.117.0`, so nothing changes for existing clients by default. Evidence that this holds: regenerating the full snapshot corpus added new `v2_117` directories and **modified no existing snapshot file**, meaning the v3/v4/v6 output is byte-for-byte unchanged. If an application supplies metadata while configured below `2.117.0`, the value is kept in memory for the local session but is neither replicated nor persisted. There is a test for this. The version-specific typebox schemas exclude the `customMetadata` field for pre-v7 formats, so no pre-v7 encoder can write it. The two formats then differ in how strictly they validate, and deliberately so: the summary schemas were already `additionalProperties: false`, so a payload falsely claiming to be pre-v7 while carrying the field is rejected there. The op `Message` schema is left permissive, as it has always been — tightening the op envelope would risk rejecting ops over envelope properties unrelated to this feature, and belongs in its own PR — so such a payload is tolerated there instead. Both behaviors are covered by tests. ## Breaking Changes None. All new API surface is `@alpha` and additive. --------- Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Copilot-Session: a037b96b-478f-4a65-9555-d7a970e7855e Copilot-Session: 42b443d7-0621-42a1-b087-f4e4765046afNoah Encke · 90337165 · 2026-08-26