De-emphasise and reduce the usage of `/_synapse` endpoints
维护者通常 14 天内回复
还没有人认领这个 Issue。
评估
- 难度
- 5/5
- 预计耗时
- 一周以上
- 新手友好度
- 25/100
- Issue 类型
- 功能
- 描述清晰度
- 需要澄清
- 活跃度
- 停滞
- 技术栈
- go
- 领域
- api, backend-api-design, testing
调研方向
首先检查 README.md 中的 Image Requirements、RegisterUser、TestCanRegisterAdmin、TestServerNotices 和 TestRegistration，以梳理当前所有 /_synapse 的使用情况。该提案没有定义替代 API，因此在 server-notice 接口和迁移范围达成一致之前，这项工作还未准备就绪；完成的标准是移除不必要的 Synapse 特定假设，同时保留 server-notice 覆盖范围。
由索引模型根据 Issue 内容生成。
描述
This is something I noticed while working on conduit support (#513), where it seems that complement currently requires or emphasizes the usage of /_synapse endpoints.
Specifically, it uses them like so:
- Under "Image Requirements" in README.md, it asks for
/_synapse/admin/v1/registerto be supported for "admin" registrations.- This is then used in
RegisterUserwhere isAdmin istrue, which is used by the following tests:TestCanRegisterAdminTestServerNotices
- This is then used in
TestRegistration, under somet.Runs
From a puritan perspective, these should not exist within complement, as they test Synapse-specific behaviour that other servers have to then copy uncritically, and without blessing or formalisation from the spec.
From such a perspective, the concept of "admins" is also dubious, as the spec (explicitly) makes no serious note of them, leaving server administration as an implementation detail. (See MSC3593 for some context reading on an attempt to broach/formalise this as well.)
For all of complement's current usage of this endpoint, there is one problem with a suggestion of removing all of them altogether; Server Notices are defined in the spec, and so the absence of a formal mechanism which empowers servers/users to send server notices is unfortunate, but this means complement has no "closed system" way of implementing this.
To make the rejection of a /_synapse specific API explicit, and to develop a better way of testing server notices, I'm suggesting to introduce a /_complement API space, which servers should enable only in their corresponding complement images, and (for now) be only used for sending server notices (exact API forthcoming), not to elevate individual users as a "server admin", as this concept does not properly exist from a matrix specification standpoint.
This API space should only perform individual elevated actions, such as sending a single server notice via a POST API.
Hopefully, this way, we can avoid biasing to a certain implementation, and help make complement more pure or agnostic when testing conformity.
- 主要语言
- Go
- 星标
- 101
- 派生
- 72
- 平均合并
- 11 天 5 小时
- 30 天内合并 PR
- 7
环境准备
- 没有 Dockerfile 或 Docker Compose 文件
- 有 Pull Request 模板
- 阅读贡献指南
从这里开始
- 先读完整个 Issue，再读项目的贡献指南。
- 在 Issue 下留言说明你要接手 —— 这能避免两个人做同样的事。
- Fork 仓库，在一个分支上完成修改。
- 提交 Pull Request，并在描述里引用这个 Issue 编号。
matrix-org/complement 的其他 Issue
-
难度 2/5 1-3 小时 新手友好度 78/100
matrix-org/complement#900 · 1 条评论 ·
维护者通常 14 天内回复
-
难度 5/5 一周以上 新手友好度 45/100
matrix-org/complement#938 ·
维护者通常 14 天内回复
-
New tests for MSC4354 Sticky Events可能重新可做 @reivilibre 于 31 天前认领，目前没有进行中的 PR。 未关闭
matrix-org/complement#925 · 已指派 1 人 ·
维护者通常 14 天内回复
-
难度 3/5 1-2 天 新手友好度 52/100
matrix-org/complement#901 · 1 条评论 ·
维护者通常 14 天内回复
-
难度 3/5 1-2 天 新手友好度 55/100
matrix-org/complement#899 ·
维护者通常 14 天内回复
查看 matrix-org/complement 的全部 Issue
相似的 Issue
-
难度 2/5 1-3 小时 新手友好度 82/100
eugenioenko/ttt#1206 ·
维护者通常 1 天内回复
-
t:urlDecodeUni does not map U+3002 and U+FF61 to a dot, unlike ModSecurity可能已有人在做 @fzipi 今天认领。 未关闭bug
难度 2/5 1-3 小时 新手友好度 78/100
维护者通常 1 天内回复
-
难度 2/5 1-3 小时 新手友好度 76/100
peasant-labs/peasant#738 ·
维护者通常 1 天内回复
-
Duplicate review still bridges two books through a bundle whose first part has a colon subtitle可能已有人在做 @vavallee 今天认领。 未关闭area:library-scan bug priority:low
难度 2/5 1-3 小时 新手友好度 83/100
维护者通常 1 天内回复
-
难度 2/5 1-3 小时 新手友好度 72/100
维护者通常 1 天内回复