docs: give release statistics a visible home on Release History - #12223
Merged
dagil-nvidia merged 5 commits intoJul 31, 2026
Merged
Conversation
github-actions
Bot
added
docs
documentation
![devin-ai-integration[bot]](https://avatars.githubusercontent.com/in/811515?s=60&v=4){kind=small}
Contributor
![dynamo-review-agent[bot]](https://avatars.githubusercontent.com/u/201626793?s=60&v=4){kind=small}
This was referenced Jul 28, 2026
dagil-nvidia
force-pushed
the
dagil-nvidia/compatibility-page-cleanup
branch
from
July 28, 2026 23:21
dac7782 to
99614d7
Compare
dagil-nvidia
force-pushed
the
dagil-nvidia/release-stats-on-history
branch
from
July 28, 2026 23:21
3780b43 to
5d8fb95
Compare
Collaborator
Author
|
/ok to test 5d8fb95 |
The per-release PR, contributor, first-time-contributor, breaking-change and known-issue counts existed only in reference/releases-data.mdx. That page is a machine-readable mirror and is deliberately hidden from the sidebar, so these counts had no reader-facing home anywhere in the docs. Every other section of that mirror already has a visible equivalent on Compatibility, Release Artifacts, Model Early Access Builds, or Release History; this was the one exception. Render them into Release History through the same marker-keyed span mechanism the support matrix uses, with the same fail-closed behavior: markers absent means the generator errors rather than appending, so placement stays under human control. Rows come from RELEASE_STATS rather than a fixed v1.0.0-v1.3.0 list, so a new release added by the per-release bump checklist shows up with no edit here. Both the visible table and the agent mirror now build from one release_stats_table helper, so they cannot drift apart -- the regenerated releases-data.mdx is byte-identical, confirming the extraction changed no output. Four rows on a short page, so a plain table rather than an accordion. Signed-off-by: Dan Gil <dagil@nvidia.com>
The "coming from" picker treated a RELEASE_STATS entry as proof that a release has per-release Deprecations and Known Issues sections to link to. That held only because RELEASE_STATS covered exactly the four v1.x stables. Gate on notesHref, which is the actual signal for a docs-native release page, so RELEASE_STATS can carry releases that are counted on Release History but have no sections to deep-link. Signed-off-by: Dan Gil <dagil@nvidia.com>
The table rendered four rows because RELEASE_STATS held four entries, not because the older releases lacked data. Backfill v0.9.0 through v0.6.0 from their GitHub bodies and record the counting rules alongside the data so later ingestions stay comparable across the two release-note eras. Pre-v1.0.0 bodies state no PR or contributor totals, so those cells stay absent and render as dashes; their first-timer lists are counted, and v0.9.0's lone Deprecation Notices entry is the same entry class the v1.x rows count under Breaking Changes. The two gaps in the existing rows are genuine: v1.0.0 states commits rather than PRs, and v1.2.0 names no first-time contributors at all. Signed-off-by: Dan Gil <dagil@nvidia.com>
Second pass at the empty prs and contributors cells, this time against the TPM release archive rather than the GitHub bodies alone. The archive corroborates the first-timer figures already published for v0.8.0 and v0.9.0 but supplies no usable PR or contributor total: its own numbers disagree with each other, and the one figure it states for v1.0.0 is a commit count that a later document reuses as a merged-PR denominator. Deriving the counts from git does not rescue it either. The closest tag-to-tag method returns 910/572/899 where the release bodies state 930/603/896 — wrong in both directions, so it is not a window offset that could be corrected for. Filling the older rows with it would put a different unit in the same column. Record all of that next to the data so the next pass starts from the evidence instead of repeating the search. Signed-off-by: Dan Gil <dagil@nvidia.com>
dagil-nvidia
force-pushed
the
dagil-nvidia/release-stats-on-history
branch
from
July 29, 2026 15:44
5d8fb95 to
603c04e
Compare
Collaborator
Author
|
/ok to test 603c04e |
Propagate #12215's main merge (0a01ca7) into #12223. Resolve conflict in docs/fern/scripts/gen_llms_tables.py PAGES dict: - Keep #12215's rename releases-data.mdx -> releases-machine-readable.mdx. - Preserve #12223's release-stats block, but relocate its page target from the legacy release-notes/README.mdx path to the current releases/release-history.mdx path so the release-stats:begin/end markers in the live Release History page continue to be updated. Signed-off-by: Daniel Gil <dagil@nvidia.com>
dagil-nvidia
added a commit
that referenced
this pull request
Jul 31, 2026
Propagate #12223's updated parent (707fac7, which itself carries #12215's merge of main and the releases-machine-readable.mdx rename) into #12241. Resolve conflict in docs/fern/pages/developer-guide/knowledge-base/ modular-components/backends/sglang/overview.md Note 1: keep this PR's corrected SGLang KV-routing text (hash-forwarding patch on Dynamo's image, fallback to text-prefix on custom builds) and update its `Source` href to the current relocated site path use-cases/multimodal-serving/ multimodal-kv-routing.md (matching notes 2/3 already using the relocated knowledge-base paths). The compatibility.mdx SGLang cell auto-merged with the same correction; the driver-floor renderer/table and generator tests carry over untouched at their current paths. Signed-off-by: Daniel Gil <dagil@nvidia.com>
dagil-nvidia
merged commit Jul 31, 2026
9934563
into
dagil-nvidia/compatibility-page-cleanup
20 checks passed
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.
Stacked on #12215. Base is
dagil-nvidia/compatibility-page-cleanup, notmain— review that PR first. Rebase ontomainand retarget once #12215 merges.Summary
The reference sidebar hides
reference/releases-data.mdx, the machine-readable release mirror. Auditing what that costs a reader turned up exactly one thing: the per-release statistics table — PRs, contributors, first-time contributors, breaking changes, known issues — had no reader-facing home anywhere in the docs. Every other section of that mirror already has a visible equivalent.This renders those counts onto Release History, so the hidden page holds nothing unique for readers.
What Changed
release-statsgenerated span onreference/release-notes/README.mdx, using the same marker-keyed mechanism as the support matrix and the same fail-closed behavior: markers absent means the generator errors rather than appending, so placement stays under human control.RELEASE_STATSrather than a fixed v1.0.0-v1.3.0 list, so a release added by the per-release bump checklist appears with no edit to the generator.release_stats_tablehelper and cannot drift. The regeneratedreleases-data.mdxis byte-identical, which confirms the extraction changed no output.README.mdx.Judgement Calls
## Release statisticsheading, matching the section name it had in the mirror so the equivalence is easy to check.-reads as zero otherwise.Verification
gen_llms_tables.py --checkclean (idempotent).check_reference.sh: ALL CHECKS PASSED, zero broken links inreference/.fern check: 0 errors.RELEASE_STATSguard exercised directly.