Tracking PR for 3.0 #105
Merged
Tracking PR for 3.0 #105
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Marwes
force-pushed
the
3.0
branch
4 times, most recently
from
c0011aa
to
f411f28
Compare
This will make mappings over parse errors be possible as desired in #86 BREAKING CHANGE Changes the type parameters of `ParseError`
Previously these streams had a `usize` which was really a pointer offset as their position which easily led to user errors. By wrapping the offset in a newtype we get custom formatting which should make it clearer that something strange is going on. As well as a simple way of converting the offset into a position. Fixes #104 BREAKING CHANGE Change `usize` to `PointerOffset` where applicable. Change `translate_position` calls into the same method on `PointerOffset`
The `range_of` parser makes it possible to "match" on the input parsed by a non-range parser and extract the range of input which was parsed. This makes range parsing much more expressive with only a small cost of an additional method on `RangeStream` Closes #83 BREAKING CHANGE `distance` must be defined on `RangeStream`
The `StreamOnce` trait do not need the position to work even if many users of the trait needs it. So this way it becomes possible for wrappers such as `State` to carry types implementing `StreamOnce` but which are not `Positioned` BREAKING CHANGE Implementors of `StreamOnce` should implement `Positioned` as well. `Positioned` must be imported if the `position` methoid is used
This makes it possible to use `choice` to create an alternation parser where the component parsers are of different types. This was still possible before by using `or` or the `choice!` macro but by overloading `choice` we get an easy to read solution without macros BREAKING CHANGE This changes the signature of the `Choice` type and the `choice` function
…ixes #79. This approach separates Positioner from the underlying Stream and from the underlying Item and Range types. This allows to track positions in streams where the Item type is third-party and the 1.0 Positioner trait cannot be implemented (due to Rust's requirement that a defining module originate one of the trait and the type). Open questions: 1) Is this abstraction correct? 2) Is it worth separating Positioner and RangePositioner? 3) Should IndexPositioner be generic over (Item, Range) or over Stream? 4) Should IndexPositioner use a single PhantomData<(Item, Range)> rather than two phantoms? 5) Is there a Rust idiom that avoids needing explicitly defining IndexPositioner::<...>::new() more frequently? Finally, if some variation of this is suitable for merging, how should we name the State type, and how should we arrange to transition the existing State code to the new world order? (I added a new module specifically to separate addition from migration.)
This lets us remove the type parameters on `IndexPositioner`
Brings byte parsers closer to char parsers
These have been superseeded by the new type and trait in the `State` module which should prove more flexible (at a slight complexity cost) BREAKING CHANGE Import and use the types from the state module instead. Simple uses should look identical
BREAKING CHANGE
…sing This makes it so that `BufferedStream::new` actually returns a `BufferedStream` value. BREAKING CHANGE Renamed used of `SharedBufferedStream` to `BufferedStream` and `BufferedStream` to `BufferedStreamRef`. Calls to `BufferedStream::new` do not change as `new` is still on `BufferedStream`
~10% when using hlists
Since it seems to be a bug preventing the no_std addition anyway it is unlikely that older versions will end up working
BREAKING CHANGE Use the recommended alternatives that has been suggested since they were depreceated (2.4.0 have all the warnings with the alternatives)
…parsers This is a tricky fix. Essentially, anytime errors are merged we lost the offset information which is used to add the correct errors after a sequencing parser has failed
…rding to or Fixes an issue with deeply nested types
This collides with the equivalent function in `combinator` but it can trivial be renamed using `recognize as recognize_range` if that is a problem.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This contains the current breaking changes I am planning to do for 3.0. More changes that are yet to be implemented are tracked in https://github.com/Marwes/combine/milestone/4.
Before this is merged I'd expect a few alpha/beta releases to be released on crates.io as well.