You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Observation-class finding, filed unassigned while working #94 (PR #140). No action requested now — this is a note that has a trigger, and the trigger is a @objectstack/cli publish.
Updated after the PM ruling on #94 (ruling: option B combined with D). A third item now sits on this page, and unlike the other two it must be removed at the trigger rather than refreshed. That asymmetry is the reason this card exists.
What is true today
#140 transcribes quickstart.mdx's Path A sample from a real boot of the published CLI, and pins the version beside the block. The reading, taken at the time:
So latest is 17.0.0 and the sample is faithful to it.
What changes at the next publish
Three claims on this page move on one event. The first two are upstream fixes that are merged but not published — both landed after 17.0.0 shipped, so neither is in what a reader installs today. The third is a caveat this repo added deliberately, whose whole purpose expires at the same moment.
The banner's Config: objectstack.config.ts row is omitted on an empty/quick-start boot. That row is in the transcribed sample today because the published CLI still prints it — beside a diagnostic that says no config was found.
os init -t app scaffolds start declaring sharingModel, so Path B's pnpm dev compiles. Path B currently exits 1 on security-owd-unset, which is why its sample block is still untranscribed and why #140 is Part of rather than a full close.
A blockquote above Path B's sample saying the sequence does not reach a dev server on @objectstack/cli 17.0.0, naming security-owd-unset and linking objectstack#9666.
⛔ DELETE — do not update
On item 3, explicitly: the admonition is a caveat with an expiry, not a fact about the page. The moment a CLI carrying objectstack#9736 publishes, it stops being true, and a warning that has stopped being true is worse than no warning — it tells a reader a working command is broken, and it does so in the house Warning: style that the corpus reserves for real hazards. Rewriting it to describe some new state would be inventing a claim nobody measured. Delete the blockquote; do not edit it, do not soften it, do not repoint it at another issue.
Suggested shape
When npm view @objectstack/cli dist-tags reports a latest above 17.0.0:
Re-boot and re-transcribe the Path A block; expect the Config: row to disappear.
Only after 1-4 does #94 close; the card is held open for exactly this.
Why file it rather than leave it in a PR body
The page now carries explicit CLI version pins, which makes staleness detectable — but only by someone who thinks to compare them against npm. Nothing sweeps for that. A card with the trigger written down is the difference between a pin being a check and a pin being decoration, which is the failure mode #94 was filed for in the first place. Item 3 sharpens that: the admonition was accepted on the condition that this card carries it, so that a temporary warning cannot quietly become permanent.
Observation-class finding, filed unassigned while working #94 (PR #140). No action requested now — this is a note that has a trigger, and the trigger is a
@objectstack/clipublish.Updated after the PM ruling on #94 (ruling: option B combined with D). A third item now sits on this page, and unlike the other two it must be removed at the trigger rather than refreshed. That asymmetry is the reason this card exists.
What is true today
#140transcribesquickstart.mdx's Path A sample from a real boot of the published CLI, and pins the version beside the block. The reading, taken at the time:So
latestis 17.0.0 and the sample is faithful to it.What changes at the next publish
Three claims on this page move on one event. The first two are upstream fixes that are merged but not published — both landed after 17.0.0 shipped, so neither is in what a reader installs today. The third is a caveat this repo added deliberately, whose whole purpose expires at the same moment.
quickstart.mdxConfig:rowConfig: objectstack.config.tsrow is omitted on an empty/quick-start boot. That row is in the transcribed sample today because the published CLI still prints it — beside a diagnostic that says no config was found.os init -t appscaffolds start declaringsharingModel, so Path B'spnpm devcompiles. Path B currently exits 1 onsecurity-owd-unset, which is why its sample block is still untranscribed and why #140 isPart ofrather than a full close.> **Warning:**admonition@objectstack/cli17.0.0, namingsecurity-owd-unsetand linking objectstack#9666.On item 3, explicitly: the admonition is a caveat with an expiry, not a fact about the page. The moment a CLI carrying objectstack#9736 publishes, it stops being true, and a warning that has stopped being true is worse than no warning — it tells a reader a working command is broken, and it does so in the house
Warning:style that the corpus reserves for real hazards. Rewriting it to describe some new state would be inventing a claim nobody measured. Delete the blockquote; do not edit it, do not soften it, do not repoint it at another issue.Suggested shape
When
npm view @objectstack/cli dist-tagsreports alatestabove 17.0.0:Config:row to disappear.MCP:must not be added to Path B unless it is measured there.Warning:blockquote outright.Only after 1-4 does #94 close; the card is held open for exactly this.
Why file it rather than leave it in a PR body
The page now carries explicit CLI version pins, which makes staleness detectable — but only by someone who thinks to compare them against npm. Nothing sweeps for that. A card with the trigger written down is the difference between a pin being a check and a pin being decoration, which is the failure mode #94 was filed for in the first place. Item 3 sharpens that: the admonition was accepted on the condition that this card carries it, so that a temporary warning cannot quietly become permanent.
Back-link: #94, #140.