buf_from_list of a non-list answers null in every state of EIGS_STRICT

未關閉
#2,467 0 則留言 0 個 reaction 已指派 0 人 在 GitHub 檢視

維護者通常 1 天內回覆

還沒有人認領這個 Issue。

評估

難度
4/5
預估耗時
3-5 天
新手友好度
58/100
Issue 類型
缺陷
描述清晰度
描述清楚
活躍度
活躍
技術堆疊
c
領域
backend

研究方向

Start with the buf_from_list builtin implementation and the strict type guard pattern used by similar builtins; then read tools/strict_differential.sh, tests/strict_guard_release_messages.tsv, and the buf_* constructor probes. Run the strict differential checks in the default and both strict modes. Done means wrong-typed inputs raise type_mismatch under EIGS_STRICT=1 without setting a variable, preserve current behavior under EIGS_STRICT=0, and the builtin documentation describes non-list behavior.

由索引模型根據 Issue 內容生成。

描述

area:stdlib kind:silent-wrong

buf_from_list given something that is not a list answers null, in every state of EIGS_STRICT, where a wrong-typed argument of a builtin raises by default (#2439):

$ cat p.eigs
try:
    x is buf_from_list of 1
    print of [x, type of x]
catch e:
    print of [e.kind, e.message]
try:
    y is buf_from_list of "a"
    print of [y, type of y]
catch e:
    print of [e.kind, e.message]
$ eigenscript p.eigs                 # phase2-numbers at f9f3e49b; the same with EIGS_STRICT=0 and EIGS_STRICT=1
[null, "null"]
[null, "null"]

The last release (v0.48.0) answers null too, so this is not from the branch: the builtin has never had a guard, and it is not among the guard probes of tools/strict_differential.sh (no row for it in tests/strict_guard_release_messages.tsv).

It is easy to reach by accident: brackets after of are an argument list, so buf_from_list of [1] passes the number 1, not a list of one. A program that writes it gets a null buffer and fails later, somewhere else.

Found beside a reading of the documents against the runtime (#1776); not one of that reading's findings.

Done when

  • buf_from_list of a value that is not a list (a number, a string, null, a dict) raises type_mismatch with no variable set and under EIGS_STRICT=1, with a message that names the builtin; under EIGS_STRICT=0 it keeps today's answer.
  • The builtin has a guard probe in tools/strict_differential.sh, in the default mode and the two others, and its row in tests/strict_guard_release_messages.tsv, or the row's absence is explained there (the release does not raise, so the release is not its oracle).
  • The other buf_* constructors that take a list or a size are probed the same way once, and each that answers quietly for a wrong type is listed on this issue (fixed here or filed).
  • docs/BUILTINS.md's row for buf_from_list says what a non-list gives.

— Claude DevBox

主要語言
C
星號
3
分支
7
平均合併
3 小時 28 分鐘
30 天內合併 PR
270

環境準備

在 Codespaces 中開啟

在瀏覽器裡用你自己的 GitHub 帳號啟動這個專案的開發容器。

從這裡開始

  1. 先讀完整個 Issue,再讀專案的貢獻指南。
  2. 在 Issue 下留言說明你要接手 —— 這能避免兩個人做同樣的事。
  3. Fork 儲存庫,在一個分支上完成修改。
  4. 送出 Pull Request,並在描述裡引用這個 Issue 編號。

InauguralSystems/EigenScript 的其他 Issue

查看 InauguralSystems/EigenScript 的全部 Issue

相似的 Issue

更多 C Issue

把新 issue 寄到你的電子郵件信箱

精選適合新手參與的 GitHub issue 摘要。