Context
After #202 collapsed the two pagination stacks, several doc comments and reference docs still
describe the pre-unification API. Each item below is a self-contained accuracy fix.
Strategy contract KDoc says [Page], but parse returns PageInfo
PaginationStrategy.parse now returns PageInfo<T> (its @return correctly says so), but two
summaries still link the removed return type:
PaginationStrategy.kt:14 — "Parses a [Response] into a [Page] ..."
PaginationStrategy.kt:37 — "Parses [response] into a [Page] of items."
The KDoc contradicts itself ([Page] in the summary vs [PageInfo] in @return), and [Page]
is now a Closeable holding a live Response — precisely what a strategy must not return —
so the stale link is actively misleading. The same phrasing sits on both paginators'
@property strategy:
Paginator.kt:81 — "Strategy that parses each response into a [Page]."
AsyncPaginator.kt:121 — "Strategy that parses each response into a [Page]."
All four should read [PageInfo].
forEachPageAsync / byPage overstate raw-body availability
The KDoc promises the delivered page's "raw Response/body is open and readable for the
duration of the callback." Every shipped strategy drains the body inside parse(...) to
materialize items, and the body is single-use, so in practice only the response metadata
(status / headers) is readable on the delivered page — not the body. The wording should say so
rather than implying the body is replayable.
implementation-plan.md WU-9 lists the removed SimplePage as shipped
The supersede banner notes "SimplePage was removed," but the "Status: shipped" paragraph just
below it (line 379) still lists "helper types SimplePage and RequestRebuilder." Drop
SimplePage from that list so the section stops contradicting itself.
refs-comparison.md still lists a separate "token" strategy and a prev_cursor
The Pagination section was only partly updated (http.paging.PagedIterable →
pagination.PagedIterable) and still says:
- "driven by a
PaginationStrategy (cursor, page-number, token, link-header)"
- "concrete impls covering cursor (
next_cursor / prev_cursor), page-number, token, and
link-header"
Only three strategies ship (Cursor / PageNumber / LinkHeader). There is no separate token
strategy — token-style APIs use CursorPaginationStrategy with a configurable query-param name,
as architecture.md already explains — and CursorPaginationStrategy is forward-only (no
prev_cursor; BiDirectionalPage is deferred). Align this section with architecture.md.
References
#202 (#30).
Context
After #202 collapsed the two pagination stacks, several doc comments and reference docs still
describe the pre-unification API. Each item below is a self-contained accuracy fix.
Strategy contract KDoc says
[Page], butparsereturnsPageInfoPaginationStrategy.parsenow returnsPageInfo<T>(its@returncorrectly says so), but twosummaries still link the removed return type:
PaginationStrategy.kt:14— "Parses a [Response] into a [Page] ..."PaginationStrategy.kt:37— "Parses [response] into a [Page] of items."The KDoc contradicts itself (
[Page]in the summary vs[PageInfo]in@return), and[Page]is now a
Closeableholding a liveResponse— precisely what a strategy must not return —so the stale link is actively misleading. The same phrasing sits on both paginators'
@property strategy:Paginator.kt:81— "Strategy that parses each response into a [Page]."AsyncPaginator.kt:121— "Strategy that parses each response into a [Page]."All four should read
[PageInfo].forEachPageAsync/byPageoverstate raw-body availabilityThe KDoc promises the delivered page's "raw
Response/body is open and readable for theduration of the callback." Every shipped strategy drains the body inside
parse(...)tomaterialize
items, and the body is single-use, so in practice only the response metadata(status / headers) is readable on the delivered page — not the body. The wording should say so
rather than implying the body is replayable.
implementation-plan.md WU-9 lists the removed
SimplePageas shippedThe supersede banner notes "
SimplePagewas removed," but the "Status: shipped" paragraph justbelow it (line 379) still lists "helper types
SimplePageandRequestRebuilder." DropSimplePagefrom that list so the section stops contradicting itself.refs-comparison.md still lists a separate "token" strategy and a
prev_cursorThe Pagination section was only partly updated (
http.paging.PagedIterable→pagination.PagedIterable) and still says:PaginationStrategy(cursor, page-number, token, link-header)"next_cursor/prev_cursor), page-number, token, andlink-header"
Only three strategies ship (
Cursor/PageNumber/LinkHeader). There is no separate tokenstrategy — token-style APIs use
CursorPaginationStrategywith a configurable query-param name,as
architecture.mdalready explains — andCursorPaginationStrategyis forward-only (noprev_cursor;BiDirectionalPageis deferred). Align this section witharchitecture.md.References
#202 (#30).