• Home
  • Features
  • Pricing
  • Docs
  • Announcements
  • Sign In

LFDT-Panurus / panurus / 35486943455
85%

Build:
DEFAULT BRANCH: main
Ran 20 Sep 2026 03:56AM UTC
Jobs 0
Files 0
Run time –
Badge
Embed ▾
README BADGES
x

If you need to use a raster PNG badge, change the '.svg' to '.png' in the link

Markdown

Textile

RDoc

HTML

Rst

pending completion
35486943455

push

github

AkramBitar
fix(storage): repair keyset pagination and guard the query DSL's landmines

Closes out the medium/low findings from the review of the SQL query-builder DSL
under token/services/storage/db/sql/query. Almost all of them are landmines
rather than live bugs: public entry points that panic or emit invalid SQL, which
no caller reaches today, so a future caller would be the one to find out.

Correctness:

- pagination.KeysetWithId shipped dead and panicking. It returned
  *keyset[I, V] for the caller's row type V, but interpreter.PreProcess and
  NewPage only type-switch on the any-erased instantiation, and generic
  instantiations are distinct concrete types, so every call fell through to
  panic in both places. The internal type collapses to keyset[I] with an
  idGetter func(any) I, so all four constructors produce the one type the
  dispatch matches.
- cond.InTuple had no empty-list guard. Both dialect interpreters
  short-circuit to writing nothing for an empty list, and And/Or can only
  filter out the AlwaysTrue/AlwaysFalse sentinels, so a condition that
  rendered to the empty string stayed in the list and emitted a leading
  operator with no left operand, e.g. "( AND status = $1)". It now returns
  AlwaysTrue on an empty field or value list, matching In/FieldIn.
- The keyset "no cursor" sentinel could collide with real data: the marker was
  the in-band value -1 (int) / "" (string), so a page whose last row
  legitimately carried one of those ids fell back to OFFSET, reintroducing the
  skip/duplicate-under-concurrent-write window keyset pagination exists to
  close. FirstID/LastID are now *I and nilElement is gone.
- WriteInTuple checked the tuples in vals only against each other and never
  against the field list they are compared with, so one field against an empty
  tuple panicked on vals[0][0]; several empty tuples rendered
  "(id) IN (($1), ($2))" while binding no parameter and desynchronising every
  later placeholder; a tuple wider than a sin... (continued)
Source Files on build 35486943455
Detailed source file information is not available for this build.
  • Back to Repo
  • Github Actions Build #35486943455
  • 35fdc251 on github
  • Prev Build on main (#35308968538)
  • Next Build on main (#35558180712)
STATUS · Troubleshooting · Open an Issue · Sales · Support · CAREERS · ENTERPRISE · START FREE TRIAL · SCHEDULE DEMO
ANNOUNCEMENTS · TWITTER · TOS & SLA · Supported CI Services · What's a CI service? · Automated Testing

© 2026 Coveralls, Inc