Navigation problems with binomial notation represented using <mmultiscripts>

未关闭
  • 哪里不对？
#823 0 条评论 0 个 reaction 已指派 0 人 在 GitHub 查看

维护者通常 1 天内回复

还没有人认领这个 Issue。

评估

难度
3/5
预计耗时
1-2 天
新手友好度
68/100
Issue 类型
缺陷
描述清晰度
基本清楚
活跃度
活跃
技术栈
rust, yaml
领域
accessibility

调研方向

Reproduce the MathML example in NVDA with MathCAT 0.7.6-rc.3, then read the relevant into-or-out-of-mmultiscripts rules in navigate.yaml and the rules that skip none and mprescripts. Done means Enhanced/Simple navigation announces semantic parts rather than literal base/subscript roles, while Character mode avoids the redundant subscript transition.

由索引模型根据 Issue 内容生成。

描述

bug rules

Navigation of a binomial coefficient represented using <mmultiscripts> produces unexpected structural labels in all three navigation modes.

MathML

<math id="0" display="block" xmlns="http://www.w3.org/1998/Math/MathML">
    <mmultiscripts id="1">
        <mi id="2">C</mi>
        <mi id="3">m</mi>
        <none id="4"/>
        <mprescripts id="5"/>
        <mi id="6">n</mi>
        <none id="7"/>
    </mmultiscripts>
</math>

This represents a binomial coefficient with presubscript n and postsubscript m.

Settings

  • Language: English
  • Speech style: SimpleSpeak
  • Speech verbosity: Medium
  • Overview: False
  • Navigation verbosity: Verbose
  • Auto zoom out: True

Tested through NVDA with MathCAT 0.7.6-rc.3 installed as an addon.

Observed behaviour

Enhanced and Simple (identical)

  1. ZoomIn

    • land on id="6"
    • hear: "zoom in; in base; n"

  2. MoveNext

    • land on id="3"
    • hear: "move right; in subscript; m"

The landing order n → m appears correct for the inferred binomial semantics, but the structural labels do not.

MathCAT's binomial-mmultiscripts inference converts this expression into a semantic binomial whose children are n and m. Therefore n is no longer the literal base of the original <mmultiscripts>, and m is no longer being navigated as its literal postsubscript.

I would expect semantic navigation to describe these as arguments/parts of the binomial rather than as "base" and "subscript".

For example, with the current generic navigation behaviour for semantic structures, something approximately like:

  1. ZoomIn

    • land on id="6"
    • hear: "zoom in; in part 1; n"

  2. MoveNext

    • land on id="3"
    • hear: "move right; in part 2; m"

would seem more consistent.

Character

  1. ZoomIn

    • land on id="2"
    • hear: "zoom in; in base; c"

  2. MoveNext

    • land on id="3"
    • hear: "move right; in subscript; m"

  3. MoveNext

    • land on id="6"
    • hear: "move right; out of subscript; in subscript; in pre-subscript; n"

Character mode correctly follows the literal MathML structure and lands on the expected nodes.

However, the third announcement appears to contain a redundant structural transition:

"out of subscript; in subscript; in pre-subscript; n"

When moving from the postsubscript m across <none> and <mprescripts> to the presubscript n, I would expect something more like:

"move right; out of subscript; in pre-subscript; n"

or another equivalent transition that does not temporarily announce "in subscript" again.

Possible causes

There appear to be two potentially relevant areas in navigate.yaml.

For Enhanced/Simple, the inferred binomial still originates from <mmultiscripts>. The into-or-out-of-mmultiscripts navigation rule applies when:

@data-from-mathml='mmultiscripts'

and assigns roles according to literal child positions (base, subscript, superscript, etc.).

This may cause the two children of the inferred semantic binomial(n, m) to be incorrectly labelled as though they were still the first two children of the original <mmultiscripts>.

For Character mode, the rules that skip <none> and <mprescripts> perform additional Move2D/MoveNextZoom operations while moving to the following real script. This may be responsible for the extra "in subscript" announcement while crossing from the postscript to the prescript.

Disclosure: I asked ChatGPT for help to dig into the cause of this, so these guesses might well be wrong.

主要语言
Rust
星标
130
派生
96
平均合并
1 天 1 小时
30 天内合并 PR
84

环境准备

  • 没有 Dockerfile 或 Docker Compose 文件
  • 没有 Pull Request 模板
  • 阅读贡献指南

从这里开始

  1. 先读完整个 Issue，再读项目的贡献指南。
  2. 在 Issue 下留言说明你要接手 —— 这能避免两个人做同样的事。
  3. Fork 仓库，在一个分支上完成修改。
  4. 提交 Pull Request，并在描述里引用这个 Issue 编号。

daisy/MathCAT 的其他 Issue

查看 daisy/MathCAT 的全部 Issue

相似的 Issue

更多 Rust Issue

把新 issue 发到你的邮箱

精选适合新手参与的 GitHub issue 摘要。