|
|
@@ -83,24 +83,26 @@ const TWO_POINT_TWO_THIRTY_TWO_CHANGELOG: ChangelogEntry = {
|
|
|
date: "2026-07-04",
|
|
|
highlights: {
|
|
|
en: [
|
|
|
- "Chat model and background-task model are now separate: the model in the chat box is only for conversation and writing prose; review, summary, de-AI, and similar tasks follow the Default Model unless you pick something else for that step.",
|
|
|
- "De-AI can use its own model, so you no longer have to share the chat model for every de-AI pass.",
|
|
|
- "Default Model moved to the top of Novel settings with clearer wording about what it actually controls.",
|
|
|
- "Model-setting labels are easier to read, and Follow default model and Follow chat model are now two separate checkboxes so they cannot mean opposite things in different places.",
|
|
|
- "Duplicate-entity detection is smoother: pick a model before scanning, last scan results are remembered, merged entries disappear right away, and the merge button no longer gets stuck doing nothing.",
|
|
|
- "Duplicate-entity scans and merges run faster, with progress that tells you what is happening; the most likely duplicates show up first.",
|
|
|
- "macOS and Linux no longer check for automatic updates — the settings page points you to manual download instead, and the install guide covers common macOS setup issues.",
|
|
|
- "Some settings pages no longer need a Save button; changes take effect as you go.",
|
|
|
+ "Chapter saving is more reliable: switching chapters no longer deletes drafts or writes to the wrong file; saving as a final chapter is when the Chinese filename is applied.",
|
|
|
+ "Chat model and background-task model are separate: the chat box is for conversation and prose; review, summary, de-AI, and similar steps use the Default Model unless you pick another for that step.",
|
|
|
+ "De-AI can use its own model instead of always sharing the chat model.",
|
|
|
+ "Duplicate-entity detection: pick a model before scanning, last results are remembered, merged entries disappear right away, and progress is clearer.",
|
|
|
+ "The editor polls disk every 2 seconds so external edits show up; fixed sync stopping after you edit locally.",
|
|
|
+ "Faster startup and safer project switching so data from the previous project does not leak in.",
|
|
|
+ "Fixed chapter title and status edits sometimes failing to save because of mismatched value types.",
|
|
|
+ "macOS and Linux no longer auto-check for updates; the settings page points you to manual download.",
|
|
|
+ "Some settings take effect immediately without a Save button.",
|
|
|
],
|
|
|
zh: [
|
|
|
- "聊天用的模型和后台任务用的模型分开了：聊天框里选的只管对话和写正文，审稿、摘要、去 AI 味等默认走「默认模型」；需要的话，也可以给单个环节单独指定模型。",
|
|
|
- "去 AI 味可以单独选一个模型，不必再和聊天模型绑在一起。",
|
|
|
- "「默认模型」提到小说设置最上面，说明也写清楚了——它负责除写正文以外的各种后台任务。",
|
|
|
- "各处的模型设置说明改写了，「跟随默认模型」和「跟随聊天模型」分开勾选，不再搞混。",
|
|
|
- "重复角色/实体检测更好用了：扫描前可以选模型，上次结果会保留，合并后条目马上消失；也修复了点击合并没反应的问题。",
|
|
|
- "重复实体扫描和合并更快了，进度会告诉你当前在干什么；最可能重复的会排在前面。",
|
|
|
- "macOS 和 Linux 版暂不支持自动更新，设置里会提示去官网手动下载；安装说明也补充了 macOS 常见问题。",
|
|
|
- "部分设置页不再需要点保存，改动即时生效。",
|
|
|
+ "章节保存更稳了：切换章节不会再误删草稿或存错文件；只有点「保存为正式章节」才会改成中文文件名。",
|
|
|
+ "聊天模型和后台任务模型分开了：聊天框里选的只管对话和写正文；审稿、摘要、去 AI 味等默认用「默认模型」，也可以给某个环节单独指定。",
|
|
|
+ "去 AI 味可以单独选模型，不用和聊天模型绑在一起。",
|
|
|
+ "重复角色/实体检测更好用：扫描前可选模型，上次结果会保留，合并后条目马上消失，进度也更清楚。",
|
|
|
+ "编辑器每 2 秒检查一次磁盘，用别的软件改了文件会自动刷新；也修好了本地改完字之后不再同步的问题。",
|
|
|
+ "启动更快，快速切换项目时不会再串上一个项目的数据。",
|
|
|
+ "修好了改章节标题或状态时，偶尔保存失败的问题。",
|
|
|
+ "macOS 和 Linux 版暂不支持软件内自动更新，设置里会提示去官网手动下载。",
|
|
|
+ "部分设置改了立刻生效，不用点保存。",
|
|
|
],
|
|
|
},
|
|
|
}
|
darknessomi