tui: hours of historical output vanish from the view while the rows are still in the DB, under a sibling session id

Ouverte
#65,244 1 commentaire 0 réactions 0 personnes assignées Voir sur GitHub

Les mainteneurs répondent en général sous 1 jour

Personne n'a encore pris cette issue.

Évaluation

Difficulté
4/5
Temps estimé
3-5 jours
Accessibilité débutants
35/100
Type d'issue
Bug
Clarté
Plutôt claire
Activité
Active
Stack technique
sqlite, typescript
Domaine
cli, database

Piste de recherche

Start in the TUI session-selection and binding code that maps the rendered view to a session id, then trace how a model/provider switch can rebind or drop that id. Use the two SQLite sessions and the session_message table as the reference: session A has the gap while session B holds the missing rows. First reproduce on a long-running TUI with a provider switch; done means the view stays bound to the original session and its full history remains visible.

Rédigé par le modèle d'indexation à partir du texte de l'issue.

Description

area/tui inference/go platform/linux triaging type/bug version/v2
Version

opencode 2.3.2 (channel latest), Linux, TUI.

What happened

While working in a single TUI, output that had been on screen for hours is no longer
visible. The current session's message history contains a hole of 21,356 s
(5 h 55 m 56 s)
— 10:56:34 -> 16:52:30 — and the turns that were rendered during
that window are not in it at all. They are still on disk, in a sibling session in the
same project and the same directory
, which is what makes this a view/binding problem
rather than data loss.

Measured (sqlite, read-only, both sessions)
session A (the one the TUI shows now) session B (holds the missing output)
id ses_e07a922f3ffe15Qu06K0bHbGU7 ses_e0763c5b5ffelNqQoTsGuwDjCv
project / directory a3b70b365cae50c572407228dc7cd6ea9ae6eb8f / /home/user same / same
model opencode-go/mimo-v2.6-flash opencode-go/glm-5.3-flash
rows (session_message) 119, 09:43:24 -> 17:21:24 453, 10:59:10 -> 16:55:21
rows inside A's hole — 451, 10:59:10 -> 16:50:20
parent_id / fork_session_id None / None None / None
time_compacting null null

Gaps in session A larger than 180 s: 307 s, 1046 s, 210 s, 21,356 s, 637 s.

Because time_compacting is null on both and neither session is a parent/fork of the
other, this is not a compaction or a fork that the user asked for — the TUI simply
stopped being bound to the conversation it was displaying.

Expected

Historical output stays visible in the TUI for the session that produced it.

Actual

The view switches to a different session id (or drops the earlier one), so hours of
output disappear from the screen even though session_message still holds every row.
The user has no way to get that transcript back from the UI.

Notes
  • Both sessions are readable and unmodified, so nothing was deleted.
  • The model differs between the two (mimo-v2.6-flash vs glm-5.3-flash), which is the
    only obvious difference; a model/provider switch around that time may be the trigger.
  • Repro steps not yet determined — observed on a long-running TUI (opened 09:43, still
    running).
Langage dominant
TypeScript
Étoiles
227k
Forks
30.7k
Merge moyen
3 h 57 min
PR mergées (30 j)
484

Préparer son environnement

Par où commencer

  1. Lisez l'issue en entier, puis le guide de contribution du projet.
  2. Signalez en commentaire que vous la prenez — cela évite que deux personnes fassent le même travail.
  3. Forkez le dépôt et travaillez sur une branche.
  4. Ouvrez une pull request qui référence le numéro de l'issue.

Autres issues de anomalyco/opencode

Toutes les issues de anomalyco/opencode

Issues similaires

Plus d'issues TypeScript

Recevez les nouvelles issues par e-mail

Un résumé court des issues GitHub adaptées aux débutants.