dots.tts 导出诊断包 · T8star-Aix

When you can't fix a dots.tts problem yourself and you're about to open a GitHub issue, T8_DotsTTS_Diagnostics is what turns "it doesn't work" into a support bundle someone can actually act on. It collects your environment, VRAM state, version info, and runtime diagnostics, scrubs out the sensitive parts, and writes a ZIP you can attach to the bug report. It's the sibling of T8_DotsTTS_Environment - that node shows you the findings as JSON, this one packages them up for a human.

The privacy story is the headline here. The node's own description is explicit about what it strips: no environment variables, no command-line arguments, no audio, no workflows, no model weights. That's a deliberate design choice, and it matters - diagnostic dumps are notorious for accidentally including API keys, token paths, and other secrets that live in environment variables. If you've ever hesitated to attach a log dump because it might contain your HF_TOKEN or your Windows username paths, this is the node that's already thought about that for you.

Mechanically it's simple. output_dir is optional - leave it blank and it writes to the pack's default project output location; set it to redirect the bundle wherever you want. The outputs are bundle_path (the absolute path to the ZIP it wrote) and report (a JSON summary of what went into it). Both are strings; wire them to text previews or just read the path off the widget.

When to reach for it

• Filing a bug report - attach the ZIP and save yourself a round of "can you run this and paste the output?" from the maintainer.
• Reproducing an issue across machines - the bundle normalizes what "your environment" means, so two people's reports are comparable.
• When a crash is environment-dependent - the same failure that works on one box and not another is usually diagnosable from this.

The one thing to remember: the bundle deliberately excludes audio and workflows, so if your bug needs a repro file, attach that separately. This node is about the environment half of the equation, and it does that half thoroughly. Installation is the pack install - ComfyUI Manager search dots-tts-t8, or clone + pip install -r requirements.txt - and it needs no model downloaded to run.