Add and document "perf" features#949
Conversation
|
I'm currently going through breaking changes and as part of that, I'm tempted to remove the |
|
@epage While you're making breaking changes, have you considered enabling simd by default, and letting people opt out of it if they really need to? I would expect far more people want the performance. |
|
Looking at #945, I was wondering if At least with I'm also split on how much defaults should be minimal vs all the bells and whistles. Two very different users who can walk away with very different negative impressions based on the result. |
That would be my reaction as well. I think there's a lot of value to defaulting to performance, rather than expecting people to opt in, when in practice the vast vast majority of people should have it enabled. |
|
On a similar note, I've been wondering if the |
Pull Request Test Coverage Report for Build 15641038728Details
💛 - Coveralls |
On the theory that if you want to write out TOML you probably want |
|
For json, a common use cases is to read and write as you are transmitting and receiving data. For machines, TOML tends to be read only. Yes, then there is also the number of ese cases left that need format preserving writes with |
|
@epage Making the |
Also point people towards "unsafe" if they want maximum performance (per toml-rs#945 showing that the features together have better performance than "perf" alone).
a06e95b to
f23f5bd
Compare
These features are not pulling there weight, only reducing parse time by about 3%. I left them on `toml_parse` for now because the API is lower level and someone may care. By doing this we remove risk and simplify the code. This is especially helpful for `toml_edit` because performance will matter less with `toml` covering more use cases. Closes toml-rs#949 BREAKING CHANGE: Removed `toml_edit` `perf` feature BREAKING CHANGE: Replaced `InternalString` with `String`
toml_parse?/simd