feat(shim): aviso "preparando" (VerbooCodar) em resposta lenta / cold start - #12
Merged
Conversation
|
Caution The consumer version of Gemini Code Assist on GitHub has been sunset. All code review activity has officially ceased. |
Mafralpm
force-pushed
the
feat/warming-idle-hint
branch
from
July 30, 2026 00:09
5036435 to
87ea13f
Compare
Quando o backend OSS esta lento (cold start ou preempcao do vLLM) e nao emite o sinal router_status:"warming", a sessao ficava parada e muda, e o usuario achava que travou (medido: glm chegou a 97s sem feedback). Agora um cronometro iniciado no ENVIO da requisicao mostra o mesmo estado de warming-up (spinner "...VerbooCodar!!") apos VERBOO_SLOW_HINT_MS (default 12s). Por iniciar no envio (e nao na leitura do corpo), cobre os dois casos: resposta lenta com a conexao ja aberta, e requisicao pendurada antes de qualquer resposta. O aviso some assim que o primeiro conteudo real chega, no fim do stream, em erro ou no abort. E so comunicacao visual: nao altera retry, nao aborta, nao muda o resultado. Testes (src/services/api/openaiShim.warmingHint.test.ts, 3 casos): 1. Corpo lento: stream que atrasa o 1o chunk. Verifica que 'warming-up' e emitido e depois limpo (null) ao chegar conteudo. 2. Requisicao pendurada: o proprio fetch demora a resolver. Verifica que 'warming-up' e emitido mesmo assim (cronometro conta desde o envio). 3. Resposta rapida: chunk chega antes do limiar. Verifica que 'warming-up' NAO e emitido (sem flicker). Regressao openaiShim*.test.ts: 117 pass. bun run build ok. tsc 0 erros novos. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
ImPedro29
force-pushed
the
feat/warming-idle-hint
branch
from
July 30, 2026 23:37
87ea13f to
d432bec
Compare
ImPedro29
approved these changes
Jul 30, 2026
ImPedro29
left a comment
There was a problem hiding this comment.
Aprovado após tornar o warming request-scoped e cobrir concorrência, chunks vazios e Responses/Codex. O cleanup agora contempla sucesso, erro, abort e stream abandonado. Validação: 201 testes focados, smoke/build e todos os checks do PR verdes.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Problema
Quando o backend fica silencioso durante cold start ou preempção, a sessão parece travada. O aviso precisa cobrir tanto fetch lento quanto atraso do primeiro conteúdo sem interferir no resultado.
Implementação
Copy
Adiciona mensagens rotativas com VerbooCodar ao spinner existente. A mudança continua sendo apenas comunicação visual: não altera retry, timeout, abort ou conteúdo da resposta.
Validação