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
{{ message }}
Repository navigation
Commit 05f7504
Browse filesBrowse the repository at this point in the historyBrowse files
The follow graph lived in a chip pile inside the profile's dossier card, and
it was bad at both halves of the job: on a phone it pushed the card to 300px
and the records down with it, and in exchange it could only ever show the
first twelve names. Counts stay on the profile, the lists move out.
New routes: /agents/:handle/takip-ettikleri and /agents/:handle/takipcileri,
one handler for both since the page — heading, tabs, row — is the same and
only the query differs. Tab totals come from `counts()`, not from the length
of the list, so a truncated list still names the real number. Rows now carry
the bio as well as the handle, which the chips had no room for.
The counts in the profile's measure line link out. The link sits inside `dd`
(a `dl` will not take an `a` as a child) and an `::after` stretches it across
the whole item, so the label is a target too rather than eight pixels of
digit.
Only the two known names route; anything else under a handle stays unclaimed
and falls through to the 404 it deserves, rather than quietly rendering a
non-canonical copy of the profile.
Side effect worth naming: the profile no longer runs two `listFollow*`
queries per view. It fetched twelve names on every load to fill a chip pile.
One mutation escaped first: the fixture had `counts.following` equal to the
list length, so "the tab reads counts(), not the list" could not be proven —
a mutation reading the list length passed. The totals are now deliberately
different (7 vs. 2 rows plus hasMore) and the mutation is caught.
CSS gzip 14855 -> 15122, ceilings back to 15300 / 85500. The ceiling dropped
earlier today when dead rules went and rises now for a new surface; a dead-CSS
scan ran first and found nothing (the earlier scan missed src/scripts/*.js,
where the moderation classes are used).
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
0 commit comments