Axel Hecht

Other file formats we use at mozilla do that (XLIFF, for example), and it's generally a PITA to deal with on the tooling side. Also, tools like Pontoon couldn't save...

I'd prefer to get rid of the structure tests, actually. That's https://github.com/projectfluent/fluent.js/issues/305, I guess. As for the reference tests, if they're structured in an unhelpful way, that'd be a good...

I referenced this from the bidi text issue, 'cause I still hope that we could do bidi and escaping in a post-format processing step, if format produced something iterable that...

In your particular use-case, is the distinction between underground and overground common enough to warrant two separate strings, each getting a positive number of floor levels? There's a level of...

I've also reproduced with nodejs nightly on WSL. I've tried a debug build on node 14, but even the `jst` macro to get the js stack trace doesn't work. I...

I've create a PR with the modification that triggers the first crash in #321. Just run the following command to reproduce: ``` node -r esm --stack-size=500000000 test/parser.js ./test/fixtures/any_char.ftl ```

I think terms should be isolated from the message. That's giving more power to localizers, in particular in scenarios where we want to have language-specific terms. Examples here would be...

One way to potentially hook this up is to add a `TONE()` function at the `Localization` class level, which you prime with metadata it needs to decide which tone to...

@spookylukey wrote a couple of interesting comments around isolation and attributes in https://github.com/django-ftl/python-fluent/blob/implement_escapers/fluent.runtime/docs/escaping.rst. I've looked at that branch in particular because I think there's some conceptual overlap between the needs...

I think we should do this, the use-cases look good enough. Localizers' life will be hard in these cases, but less hard than with the alternative.