# Description
I'm on rust toolchain 1.8.4, and I can see clippy warnings that can't be
caught by the ci workflow, primarily related to lifetime params.
I think it doesn't hurt to fix those in advance.
# User-Facing Changes
# Tests + Formatting
# After Submitting
This is good practice as all our iterators will never return a value
after reaching `None`
The benefit should be minimal as only `Iterator::fuse` is directly
specialized and itself rarely used (sometimes in `itertools` adaptors)
Thus it is mostly a documentation thing
# Description
This adds a `SharedCow` type as a transparent copy-on-write pointer that
clones to unique on mutate.
As an initial test, the `Record` within `Value::Record` is shared.
There are some pretty big wins for performance. I'll post benchmark
results in a comment. The biggest winner is nested access, as that would
have cloned the records for each cell path follow before and it doesn't
have to anymore.
The reusability of the `SharedCow` type is nice and I think it could be
used to clean up the previous work I did with `Arc` in `EngineState`.
It's meant to be a mostly transparent clone-on-write that just clones on
`.to_mut()` or `.into_owned()` if there are actually multiple
references, but avoids cloning if the reference is unique.
# User-Facing Changes
- `Value::Record` field is a different type (plugin authors)
# Tests + Formatting
- 🟢 `toolkit fmt`
- 🟢 `toolkit clippy`
- 🟢 `toolkit test`
- 🟢 `toolkit test stdlib`
# After Submitting
- [ ] use for `EngineState`
- [ ] use for `Value::List`
# Description
This decouples the serialized representation of `Record` from its
internal implementation. It now gets treated as a map type in `serde`.
This has several benefits:
- more efficient representation (not showing inner fields)
- human readable e.g. as a JSON record
- no breaking changes when refactoring the `Record` internals in the
future (see #12326, or potential introduction of `indexmap::IndexMap`
for large N)
- we now deny the creation of invalid records a non-cooperating plugin
could produce
- guaranteed key-value correspondence
- checking for unique keys
# Breaking change to the plugin protocol:
Now expects a record/map directly as the `Record.val` field instead of a
serialization of it.
# Description
Requested by @ayax79. This makes the custom value behavior more correct,
by calling the methods on the plugin to handle the custom values in
examples rather than the methods on the custom values themselves. This
helps for handle-type custom values (like what he's doing with
dataframes).
- Equality checking in `PluginTest::test_examples()` changed to use
`PluginInterface::custom_value_partial_cmp()`
- Base value rendering for `PluginSignature` changed to use
`Plugin::custom_value_to_base_value()`
- Had to be moved closer to `serve_plugin` for this reason, so the test
for writing signatures containing custom values was removed
- That behavior should still be tested to some degree, since if custom
values are not handled, signatures will fail to parse, so all of the
other tests won't work.
# User-Facing Changes
- `Record::sort_cols()` method added to share functionality required by
`PartialCmp`, and it might also be slightly faster
- Otherwise, everything should mostly be the same but better. Plugins
that don't implement special handling for custom values will still work
the same way, because the default implementation is just a pass-through
to the `CustomValue` methods.
# Tests + Formatting
- 🟢 `toolkit fmt`
- 🟢 `toolkit clippy`
- 🟢 `toolkit test`
- 🟢 `toolkit test stdlib`
# Description
This shrinks `Record`'s size in half and and allows you to include it in
`Value` without growing the size.
Changing the `Record` internals may have slightly different performance
characteristics as the cache locality changes on lookups (if you
directly need the value, it should be closer, but in other cases may
blow up the cache line budget)
Also different perf characteristics on creation expected.
`Record::from_raw_cols_vals` now probably worse.
## Benchmarking
Comparison with the main branch (boxed Record) revealed no significant
change to the creation but an improvement when accessing larger N.
The fact that this was more pronounced for nested access (still cloning
before nushell/nushell#12325) leads to the conclusion that this may
still be dominated by the smaller clone necessary for a 24-byte `Record`
over the previous 48 bytes.
# User-Facing Changes
Reduced memory usage
# Description
Does a little cleanup in `record.rs`:
- Makes the `record!` macro more hygienic.
- Converts regular comments to doc comments from #11718.
- Converts the `Record` iterators to new types.
<!--
if this PR closes one or more issues, you can automatically link the PR
with
them by using one of the [*linking
keywords*](https://docs.github.com/en/issues/tracking-your-work-with-issues/linking-a-pull-request-to-an-issue#linking-a-pull-request-to-an-issue-using-a-keyword),
e.g.
- this PR should close #xxxx
- fixes #xxxx
you can also mention related issues, PRs or discussions!
-->
# Description
<!--
Thank you for improving Nushell. Please, check our [contributing
guide](../CONTRIBUTING.md) and talk to the core team before making major
changes.
Description of your pull request goes here. **Provide examples and/or
screenshots** if your changes affect the user experience.
-->
Fixes https://github.com/nushell/nushell/issues/11716
The problem is in our [record creation
API](0d518bf813/crates/nu-protocol/src/value/record.rs (L33))
which panics if the numbers of columns and values are different. I added
a safe variant that returns a `Result` and used it in the `rotate`
command.
## TODO in another PR:
Go through all `from_raw_cols_vals_unchecked()` (this includes the
`record!` macro which uses the unchecked version) and make sure that
either
a) it is guaranteed the number of cols and vals is the same, or
b) convert the call to `from_raw_cols_vals()`
Reason: Nushell should never panic.
# User-Facing Changes
<!-- List of all changes that impact the user experience here. This
helps us keep track of breaking changes. -->
# Tests + Formatting
<!--
Don't forget to add tests that cover your changes.
Make sure you've run and fixed any issues with these commands:
- `cargo fmt --all -- --check` to check standard code formatting (`cargo
fmt --all` applies these changes)
- `cargo clippy --workspace -- -D warnings -D clippy::unwrap_used` to
check that you're using the standard code style
- `cargo test --workspace` to check that all tests pass (on Windows make
sure to [enable developer
mode](https://learn.microsoft.com/en-us/windows/apps/get-started/developer-mode-features-and-debugging))
- `cargo run -- -c "use std testing; testing run-tests --path
crates/nu-std"` to run the tests for the standard library
> **Note**
> from `nushell` you can also use the `toolkit` as follows
> ```bash
> use toolkit.nu # or use an `env_change` hook to activate it
automatically
> toolkit check pr
> ```
-->
# After Submitting
<!-- If your PR had any user-facing changes, update [the
documentation](https://github.com/nushell/nushell.github.io) after the
PR is merged, if necessary. This will help us keep the docs up to date.
-->
# Description
With #11386 we don't have any nushell-internal code directly accessing
the `vals` field of `Record`, so let's make it private so everyone in
the future uses the checked ways guaranteeing matching cols/vals.
The `cols` feel has to remain pub for now as `rename` still directly
mutates this field. See #11020 for challenges for this refactor.
# Plugin-Author-Facing Changes
This is a breaking change for outside plugins that relied on the `pub`
fields.
# Description
Compatible with `Vec::truncate` and `indexmap::IndexMap::truncate`
Found useful in #10903 for `drop column`
# Tests + Formatting
Doctest with the relevant edge-cases
Matches the general behavior of `Vec::drain` or
`indexmap::IndexMap::drain`:
- Drop the remaining elements (implementing the unstable `keep_rest()`
would not be compatible with something like `indexmap`)
- No `AsRef<[T]>` or `Drain::as_slice()` behavior as this would make
layout assumptions.
- `Drain: DoubleEndedIterator`
Found useful in #10903
# Description
Replaces the `Vec::remove` in `Record::retain_mut` with some swaps which
should eliminate the `O(n^2)` complexity due to repeated shifting of
elements.
# Description
While we have now a few ways to add items or iterate over the
collection, we don't have a way to cleanly remove items from `Record`.
This PR fixes that:
- Add `Record.remove()` to remove by key
- makes the assumption that keys are unique, so can not be used
universally, yet (see #10875 for an important example)
- Add naive `Record.retain()` for inplace removal
- This follows the two separate `retain`/`retain_mut` in the Rust std
library types, compared to the value-mutating `retain` in `indexmap`
- Add `Record.retain_mut()` for one-pass pruning
Continuation of #10841
# User-Facing Changes
None yet.
# Tests + Formatting
Doctests for the `retain`ing fun
# Description
> Our `Record` looks like a map, quacks like a map, so let's treat it
with the API for a map
Implement common methods found on e.g. `std::collections::HashMap` or
the insertion-ordered [indexmap](https://docs.rs/indexmap).
This allows contributors to not have to worry about how to get to the
relevant items and not mess up the assumptions of a Nushell record.
## Record assumptions
- `cols` and `vals` are of equal length
- for all practical purposes, keys/columns should be unique
## End goal
The end goal of the upcoming series of PR's is to allow us to make
`cols` and `vals` private.
Then it would be possible to exchange the backing datastructure to best
fit the expected workload.
This could be statically (by finding the best balance) or dynamically by
using an `enum` of potential representations.
## Parts
- Add validating explicit part constructor
`Record::from_raw_cols_vals()`
- Add `Record.columns()` iterator
- Add `Record.values()` iterator
- Add consuming `Record.into_values()` iterator
- Add `Record.contains()` helper
- Add `Record.insert()` that respects existing keys
- Add key-based `.get()`/`.get_mut()` to `Record`
- Add `Record.get_index()` for index-based access
- Implement `Extend` for `Record` naively
- Use checked constructor in `record!` macro
- Add `Record.index_of()` to get index by key
# User-Facing Changes
None directly
# Developer facing changes
You don't have to roll your own record handling and can use a familiar
API
# Tests + Formatting
No explicit unit tests yet. Wouldn't be too tricky to validate core
properties directly.
Will be exercised by the following PRs using the new
methods/traits/iterators.
# Description
This PR creates a new `Record` type to reduce duplicate code and
possibly bugs as well. (This is an edited version of #9648.)
- `Record` implements `FromIterator` and `IntoIterator` and so can be
iterated over or collected into. For example, this helps with
conversions to and from (hash)maps. (Also, no more
`cols.iter().zip(vals)`!)
- `Record` has a `push(col, val)` function to help insure that the
number of columns is equal to the number of values. I caught a few
potential bugs thanks to this (e.g. in the `ls` command).
- Finally, this PR also adds a `record!` macro that helps simplify
record creation. It is used like so:
```rust
record! {
"key1" => some_value,
"key2" => Value::string("text", span),
"key3" => Value::int(optional_int.unwrap_or(0), span),
"key4" => Value::bool(config.setting, span),
}
```
Since macros hinder formatting, etc., the right hand side values should
be relatively short and sweet like the examples above.
Where possible, prefer `record!` or `.collect()` on an iterator instead
of multiple `Record::push`s, since the first two automatically set the
record capacity and do less work overall.
# User-Facing Changes
Besides the changes in `nu-protocol` the only other breaking changes are
to `nu-table::{ExpandedTable::build_map, JustTable::kv_table}`.
