Better handling of whitespace in serialization #31
Conversation
* Update tested node versions and remove 0.12 * Update linting and run eslint --fix on commit * Use correct check to serialize NaN
* Remove leftover .only
|
Ok, I've added a few commits here myself as I wanted to fix the Node 0.12 test issue. For which I found the best solution was to stop testing on Node 0.12 but instead align with rollup in testing Node 4, 6, 8 (and 10). While doing this, I decided to add a prepare script and fix the long ignored linting to actually run (with the --fix option to auto-correct annoying white-space issues) on commit. |
|
There was also in issue with serializing NaN that actually the linting discovered (NaN !== NaN...). I've added this to the test and fixed it. |
|
Thanks for checking that and adding the changes! The reason we can't use JSON.stringify is for serialization of dates and regexes needed for Toml or Yaml I think, so nice to get a single implementation that covers both. Then also useful to have legal keys unquoted for brevity. |
That does indeed look very nice! |
Yes, I rewrote the wheel, but it's in the name of supporting custom types and non-unnecessary quotes, and I really couldn't find anything that did everything we need here.