<!--
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.
-->
This PR adds type checking of all command input types at run-time.
Generally, these errors should be caught by the parser, but sometimes we
can't know the type of a value at parse-time. The simplest example is
using the `echo` command, which has an output type of `any`, so
prefixing a literal with `echo` will bypass parse-time type checking.
Before this PR, each command has to individually check its input types.
This can result in scenarios where the input/output types don't match
the actual command behavior. This can cause valid usage with an
non-`any` type to become a parse-time error if a command is missing that
type in its pipeline input/output (`drop nth` and `history import` do
this before this PR). Alternatively, a command may not list a type in
its input/output types, but doesn't actually reject that type in its
code, which can have unintended side effects (`get` does this on an
empty pipeline input, and `sort` used to before #13154).
After this PR, the type of the pipeline input is checked to ensure it
matches one of the input types listed in the proceeding command's
input/output types. While each of the issues in the "before this PR"
section could be addressed with each command individually, this PR
solves this issue for _all_ commands.
**This will likely cause some breakage**, as some commands have
incorrect input/output types, and should be adjusted. Also, some scripts
may have erroneous usage of commands. In writing this PR, I discovered
that `toolkit.nu` was passing `null` values to `str join`, which doesn't
accept nothing types (if folks think it should, we can adjust it in this
PR or in a different PR). I found some issues in the standard library
and its tests. I also found that carapace's vendor script had an
incorrect chaining of `get -i`:
```nushell
let expanded_alias = (scope aliases | where name == $spans.0 | get -i 0 | get -i expansion)
```
Before this PR, if the `get -i 0` ever actually did evaluate to `null`,
the second `get` invocation would error since `get` doesn't operate on
`null` values. After this PR, this is immediately a run-time error,
alerting the user to the problematic code. As a side note, we'll need to
PR this fix (`get -i 0 | get -i expansion` -> `get -i 0.expansion`) to
carapace.
A notable exception to the type checking is commands with input type of
`nothing -> <type>`. In this case, any input type is allowed. This
allows piping values into the command without an error being thrown. For
example, `123 | echo $in` would be an error without this exception.
Additionally, custom types bypass type checking (I believe this also
happens during parsing, but not certain)
I added a `is_subtype` method to `Value` and `PipelineData`. It
functions slightly differently than `get_type().is_subtype()`, as noted
in the doccomments. Notably, it respects structural typing of lists and
tables. For example, the type of a value `[{a: 123} {a: 456, b: 789}]`
is a subtype of `table<a: int>`, whereas the type returned by
`Value::get_type` is a `list<any>`. Similarly, `PipelineData` has some
special handling for `ListStream`s and `ByteStream`s. The latter was
needed for this PR to work properly with external commands.
Here's some examples.
Before:
```nu
1..2 | drop nth 1
Error: nu::parser::input_type_mismatch
× Command does not support range input.
╭─[entry #9:1:8]
1 │ 1..2 | drop nth 1
· ────┬───
· ╰── command doesn't support range input
╰────
echo 1..2 | drop nth 1
# => ╭───┬───╮
# => │ 0 │ 1 │
# => ╰───┴───╯
```
After this PR, I've adjusted `drop nth`'s input/output types to accept
range input.
Before this PR, zip accepted any value despite not being listed in its
input/output types. This caused different behavior depending on if you
triggered a parse error or not:
```nushell
1 | zip [2]
# => Error: nu::parser::input_type_mismatch
# =>
# => × Command does not support int input.
# => ╭─[entry #3:1:5]
# => 1 │ 1 | zip [2]
# => · ─┬─
# => · ╰── command doesn't support int input
# => ╰────
echo 1 | zip [2]
# => ╭───┬───────────╮
# => │ 0 │ ╭───┬───╮ │
# => │ │ │ 0 │ 1 │ │
# => │ │ │ 1 │ 2 │ │
# => │ │ ╰───┴───╯ │
# => ╰───┴───────────╯
```
After this PR, it works the same in both cases. For cases like this, if
we do decide we want `zip` or other commands to accept any input value,
then we should explicitly add that to the input types.
```nushell
1 | zip [2]
# => Error: nu::parser::input_type_mismatch
# =>
# => × Command does not support int input.
# => ╭─[entry #3:1:5]
# => 1 │ 1 | zip [2]
# => · ─┬─
# => · ╰── command doesn't support int input
# => ╰────
echo 1 | zip [2]
# => Error: nu:🐚:only_supports_this_input_type
# =>
# => × Input type not supported.
# => ╭─[entry #14:2:6]
# => 2 │ echo 1 | zip [2]
# => · ┬ ─┬─
# => · │ ╰── only list<any> and range input data is supported
# => · ╰── input type: int
# => ╰────
```
# User-Facing Changes
<!-- List of all changes that impact the user experience here. This
helps us keep track of breaking changes. -->
**Breaking change**: The type of a command's input is now checked
against the input/output types of that command at run-time. While these
errors should mostly be caught at parse-time, in cases where they can't
be detected at parse-time they will be caught at run-time instead. This
applies to both internal commands and custom commands.
Example function and corresponding parse-time error (same before and
after PR):
```nushell
def foo []: int -> nothing {
print $"my cool int is ($in)"
}
1 | foo
# => my cool int is 1
"evil string" | foo
# => Error: nu::parser::input_type_mismatch
# =>
# => × Command does not support string input.
# => ╭─[entry #16:1:17]
# => 1 │ "evil string" | foo
# => · ─┬─
# => · ╰── command doesn't support string input
# => ╰────
# =>
```
Before:
```nu
echo "evil string" | foo
# => my cool int is evil string
```
After:
```nu
echo "evil string" | foo
# => Error: nu:🐚:only_supports_this_input_type
# =>
# => × Input type not supported.
# => ╭─[entry #17:1:6]
# => 1 │ echo "evil string" | foo
# => · ──────┬────── ─┬─
# => · │ ╰── only int input data is supported
# => · ╰── input type: string
# => ╰────
```
Known affected internal commands which erroneously accepted any type:
* `str join`
* `zip`
* `reduce`
# 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 toolkit.nu; toolkit test stdlib"` 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
> ```
-->
- 🟢 `toolkit fmt`
- 🟢 `toolkit clippy`
- 🟢 `toolkit test`
- 🟢 `toolkit test stdlib`
# 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.
-->
* Play whack-a-mole with the commands and scripts this will inevitably
break
# Description
Currently the step size of range values are discarded when converting to
nuon. This PR fixes that and makes `to nuon | from nuon` round trips
work.
# User-Facing Changes
`to nuon` conversion of `range` values now include the step size
# Tests + Formatting
Added some additional tests to cover inclusive/exclusive integer/float
and step size cases.
# Description
This PR introduces a switch `--serialize` that allows serializing of
types that cannot be deserialized. Right now it only serializes closures
as strings in `to toml`, `to json`, `to nuon`, `to text`, some indirect
`to html` and `to yaml`.
A lot of the changes are just weaving the engine_state through calling
functions and the rest is just repetitive way of getting the closure
block span and grabbing the span's text.
In places where it has to report `<Closure 123>` I changed it to
`closure_123`. It always seemed like the `<>` were not very nushell-y.
This is still a breaking change.
I think this could also help with systematic translation of old config
to new config file.
# 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 toolkit.nu; toolkit test stdlib"` 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.
-->
<!--
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.
-->
I realized that the `into bool` command somehow implements a conversion
into a boolean value which was very similar to my implementation of
~`Value::as_env_bool`~ `Value::coerce_bool`. To streamline that behavior
a bit, I replaced most of the implementation of `into bool` with my
~`Value::as_env_bool`~ `Value::coerce_bool` method.
Also I added a new flag called `--relaxed` which lets the command behave
more closely to the ~`Value::as_env_bool`~ `Value::coerce_bool` method
as it allows null values and is more loose to strings. ~Which now begs
the question, should I rename `Value::as_env_bool` just to
`Value::coerce_bool` which would fit the `Value::coerce_str` method
name?~ (Renamed that.)
# User-Facing Changes
<!-- List of all changes that impact the user experience here. This
helps us keep track of breaking changes. -->
The `into bool` command behaves the same but with `--relaxed` you can
also throw a `null` or some more strings at it which makes it more
ergonomic for env conversions.
# 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 toolkit.nu; toolkit test stdlib"` 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
> ```
-->
I added some more tests to see that the strict handling works and added
some more examples to the command to showcase the `--relaxed` flag which
also gets tested.
- 🟢 `toolkit fmt`
- 🟢 `toolkit clippy`
- 🟢 `toolkit test`
- 🟢 `toolkit test stdlib`
# 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.
-->
@Bahex mentioned in #14704 that it broke the zoxide script, this PR
should help to fix the issue.
<!--
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.
-->
The `Value::coerce_str` method weirdly doesn't allow coercing boolean
values into strings while commands like `true | into string` work
without issues. So I added that.
# User-Facing Changes
<!-- List of all changes that impact the user experience here. This
helps us keep track of breaking changes. -->
This is technically a breaking change if a nushell library user depended
on the fact that boolean values weren't coerceable to strings. But I
doubt that really.
# 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 toolkit.nu; toolkit test stdlib"` 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
> ```
-->
- 🟢 `toolkit fmt`
- 🟢 `toolkit clippy`
- 🟢 `toolkit test`
- 🟢 `toolkit test stdlib`
Following #14700 we should make sure more folks are aware that you
shouldn't use `internal_span` outside of `Value` or core protocol/engine
internals.
By making it a doccomment maybe a few folks see the text in the lsp
hover etc.
<!--
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.
-->
In this PR I continued the idea of #11494, it added an `auto` option to
the ansi coloring config option, I did this too but in a more simple
approach.
So I added a new enum `UseAnsiColoring` with the three values `True`,
`False` and `Auto`. When that value is set to `auto`, the default value,
it will use `std::io::stdout().is_terminal()` to decided whether to use
ansi coloring. This allows to dynamically decide whether to print ansi
color codes or not, [cargo does it the same
way](652623b779/src/bin/cargo/main.rs (L72)).
`True` and `False` act as overrides to the `is_terminal` check. So with
that PR it is possible to force ansi colors on the `table` command or
automatically remove them from the miette errors if no terminal is used.
# User-Facing Changes
<!-- List of all changes that impact the user experience here. This
helps us keep track of breaking changes. -->
Terminal users shouldn't be affected by this change as the default value
was `true` and `is_terminal` returns for terminals `true` (duh).
Non-terminal users, that use `nu` in some embedded way or the engine
implemented in some other way (like my jupyter kernel) will now have by
default no ansi coloring and need to enable it manually if their
environment allows it.
# 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 toolkit.nu; toolkit test stdlib"` 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
> ```
-->
The test for fancy errors expected ansi codes, since tests aren't run
"in terminal", the ansi codes got stripped away.
I added a line that forced ansi colors above it. I'm not sure if that
should be the case or if we should test against no ansi colors.
- 🟢 `toolkit fmt`
- 🟢 `toolkit clippy`
- 🟢 `toolkit test`
- 🟢 `toolkit test stdlib`
# 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.
-->
This should resolve#11464 and partially #11847. This also closes
#11494.
# Description
Adds a new `Filesize` type so that `FromValue` can be used to convert a
`Value::Filesize` to a `Filesize`. Currently, to extract a filesize from
a `Value` using `FromValue`, you have to extract an `i64` which coerces
`Value::Int`, `Value::Duration`, and `Value::Filesize` to an `i64`.
Having a separate type also allows us to enforce checked math to catch
overflows. Similarly, it allows us to specify other trait
implementations like `Display` in a common place.
# User-Facing Changes
Multiplication with filesizes now error on overflow. Should not be a
breaking change for plugins (i.e., serialization) since `Filesize` is
marked with `serde(transparent)`.
# Tests + Formatting
Updated some tests.
# Description
The "append" operator currently serves as both the append operator and
the concatenation operator. This dual role creates ambiguity when
operating on nested lists.
```nu
[1 2] ++ 3 # appends a value to a list [1 2 3]
[1 2] ++ [3 4] # concatenates two lists [1 2 3 4]
[[1 2] [3 4]] ++ [5 6]
# does this give [[1 2] [3 4] [5 6]]
# or [[1 2] [3 4] 5 6]
```
Another problem is that `++=` can change the type of a variable:
```nu
mut str = 'hello '
$str ++= ['world']
($str | describe) == list<string>
```
Note that appending is only relevant for lists, but concatenation is
relevant for lists, strings, and binary values. Additionally, appending
can be expressed in terms of concatenation (see example below). So, this
PR changes the `++` operator to only perform concatenation.
# User-Facing Changes
Using the `++` operator with a list and a non-list value will now be a
compile time or runtime error.
```nu
mut list = []
$list ++= 1 # error
```
Instead, concatenate a list with one element:
```nu
$list ++= [1]
```
Or use `append`:
```nu
$list = $list | append 1
```
# After Submitting
Update book and docs.
---------
Co-authored-by: Douglas <32344964+NotTheDr01ds@users.noreply.github.com>
# Description
Fixes#14294 - Turned out to be a whole lot easier than I expected, but
please double-check me on this, since it's an area I haven't been in
before.
# User-Facing Changes
Allow date to be added to a duration type.
# Tests + Formatting
Tests added:
* Duration + Date is allowed
* Duration - Date is not allowed
# Description
Turns out there are duplicate conversion functions: `as_i64` and
`as_f64`. In most cases, these can be replaced with `as_int` and
`as_float`, respectively.
# Description
Dividing two ints can currently return either an int or a float. Not
having a single return type for an operation between two types seems
problematic. Additionally, the type signature for division says that
dividing two ints returns only an int which does not match the current
implementation (it can also return a float). This PR changes division
between almost all types to return a float (except for `filesize /
number` or `duration / number`, since there are no float representations
for these types).
Currently, floor division between certain types is not implemented even
though the type signature allows it. Also, the current implementation of
floor division uses a combination of clamping and flooring rather than
simply performing floor division which this PR fixes. Additionally, the
signature was changed so that `int // float`, `float // int`, and `float
// float` now return float instead of int. This matches the automatic
float promotion in the rest of the operators (as well as how Python does
floor division which I think is the original inspiration).
Since regular division has always returned fractional values (and now
returns a float to reflect that), `mod` is now defined in terms of floor
division. That is, `D // d = q`, `D mod d = r`, and `D = d * q + r `.
This is just like the `%` operator in Python, which is also based off
floor division (at least for ints and floats). Additionally,
implementations missing from `mod`'s current type signature have been
added (`duration mod int` and `duration mod float`).
This PR also overhauls the overflow checking and errors for div, mod,
and floor div. If an operation overflows, it will now cause an error.
# User-Facing Changes
- Div now returns a float in most cases.
- Floor division now actually does floor division.
- Floor division now does automatic float promotion, returning a float
in more instances.
- Floor division now actually allows division with filesize and
durations as its type signature claimed.
- Mod is now defined and implemented in terms of floor division rather
than truncating division.
- Mod now actually allows filesize and durations as its type signature
claimed.
- Div, mod, and floor div now all have proper overflow checks.
## Examples
When the divisor and the dividend have the same sign, the quotient and
remainder will be the same as before. (Except that this PR will give
more accurate results, since it does not do an intermediate float
conversion). If the signs of the divisor and dividend are different,
then the results will be different, or rather actually correct.
Before:
```nu
let q = 8 // -3 # -3
let r = 8 mod -3 # 2
8 == $q * -3 + $r # false
```
After:
```nu
let q = 8 // -3 # -3
let r = 8 mod -3 # -1
8 == $q * -3 + $r # true
```
Before:
```nu
let q = -8 // 3 # -3
let r = -8 mod 3 # -2
-8 == $q * 3 + $r # false
```
After:
```nu
let q = -8 // 3 # -3
let r = -8 mod 3 # 1
-8 == $q * 3 + $r # true
```
# Tests + Formatting
Added a few tests.
# After Submitting
Probably update the docs.
# Description
This PR changes the range contains logic to take the step into account.
```nushell
# before
2 in 1..3.. # true
# now
2 in 1..3.. # false
```
---
I encountered another issue while adding tests. Due to floating point
precision, `2.1 in 1..1.1..3` will return `false`. The floating point
error is even bigger than `f64::EPSILON` (`0.09999999999999876` vs
`2.220446049250313e-16`). This issue disappears with bigger numbers.
I tried a different algorithm (checking if the estimated number of steps
is close enough to any integer) but the results are still pretty bad:
```rust
let n_steps = (value - self.start) / self.step; // 14.999999999999988
(n_steps - n_steps.round()).abs() < f64::EPSILON // returns false
```
Maybe it can be shipped like this, the REPL already has floating point
errors (`1.1 - 1` returns `0.10000000000000009`). Or maybe there's a way
to fix this that I didn't think of. I'm open to ideas! But in any case
performing this kind of checks on a range of floats seems more niche
than doing it on a range of ints.
# User-Facing Changes
Code that depended on this behavior to check if a number is between
`start` and `end` will potentially return a different value.
# Tests + Formatting
# After Submitting
# Description
Closes#12535
Implements sort-by functionality of #8322
Fixes sort-by part of #8667
This PR does two main things: add a new cell path and closure parameter
to `sort-by`, and attempt to make Nushell's sorting behavior
well-defined.
## `sort-by` features
The `columns` parameter is replaced with a `comparator` parameter, which
can be a cell path or a closure. Examples are from docs PR.
1. Cell paths
The basic interactive usage of `sort-by` is the same. For example, `ls |
sort-by modified` still works the same as before. It is not quite a
drop-in replacement, see [behavior changes](#behavior-changes).
Here's an example of how the cell path comparator might be useful:
```nu
> let cities = [
{name: 'New York', info: { established: 1624, population: 18_819_000 } }
{name: 'Kyoto', info: { established: 794, population: 37_468_000 } }
{name: 'São Paulo', info: { established: 1554, population: 21_650_000 }
}
]
> $cities | sort-by info.established
╭───┬───────────┬────────────────────────────╮
│ # │ name │ info │
├───┼───────────┼────────────────────────────┤
│ 0 │ Kyoto │ ╭─────────────┬──────────╮ │
│ │ │ │ established │ 794 │ │
│ │ │ │ population │ 37468000 │ │
│ │ │ ╰─────────────┴──────────╯ │
│ 1 │ São Paulo │ ╭─────────────┬──────────╮ │
│ │ │ │ established │ 1554 │ │
│ │ │ │ population │ 21650000 │ │
│ │ │ ╰─────────────┴──────────╯ │
│ 2 │ New York │ ╭─────────────┬──────────╮ │
│ │ │ │ established │ 1624 │ │
│ │ │ │ population │ 18819000 │ │
│ │ │ ╰─────────────┴──────────╯ │
╰───┴───────────┴────────────────────────────╯
```
2. Key closures
You can supply a closure which will transform each value into a sorting
key (without changing the underlying data). Here's an example of a key
closure, where we want to sort a list of assignments by their average
grade:
```nu
> let assignments = [
{name: 'Homework 1', grades: [97 89 86 92 89] }
{name: 'Homework 2', grades: [91 100 60 82 91] }
{name: 'Exam 1', grades: [78 88 78 53 90] }
{name: 'Project', grades: [92 81 82 84 83] }
]
> $assignments | sort-by { get grades | math avg }
╭───┬────────────┬───────────────────────╮
│ # │ name │ grades │
├───┼────────────┼───────────────────────┤
│ 0 │ Exam 1 │ [78, 88, 78, 53, 90] │
│ 1 │ Project │ [92, 81, 82, 84, 83] │
│ 2 │ Homework 2 │ [91, 100, 60, 82, 91] │
│ 3 │ Homework 1 │ [97, 89, 86, 92, 89] │
╰───┴────────────┴───────────────────────╯
```
3. Custom sort closure
The `--custom`, or `-c`, flag will tell `sort-by` to interpret closures
as custom sort closures. A custom sort closure has two parameters, and
returns a boolean. The closure should return `true` if the first
parameter comes _before_ the second parameter in the sort order.
For a simple example, we could rewrite a cell path sort as a custom sort
(see
[here](https://github.com/nushell/nushell.github.io/pull/1568/files#diff-a7a233e66a361d8665caf3887eb71d4288000001f401670c72b95cc23a948e86R231)
for a more complex example):
```nu
> ls | sort-by -c {|a, b| $a.size < $b.size }
╭───┬─────────────────────┬──────┬──────────┬────────────────╮
│ # │ name │ type │ size │ modified │
├───┼─────────────────────┼──────┼──────────┼────────────────┤
│ 0 │ my-secret-plans.txt │ file │ 100 B │ 10 minutes ago │
│ 1 │ shopping_list.txt │ file │ 100 B │ 2 months ago │
│ 2 │ myscript.nu │ file │ 1.1 KiB │ 2 weeks ago │
│ 3 │ bigfile.img │ file │ 10.0 MiB │ 3 weeks ago │
╰───┴─────────────────────┴──────┴──────────┴────────────────╯
```
## Making sort more consistent
I think it's important for something as essential as `sort` to have
well-defined semantics. This PR contains some changes to try to make the
behavior of `sort` and `sort-by` consistent. In addition, after working
with the internals of sorting code, I have a much deeper understanding
of all of the edge cases. Here is my attempt to try to better define
some of the semantics of sorting (if you are just interested in changes,
skip to "User-Facing changes")
- `sort`, `sort -v`, and `sort-by` now all work the same. Each
individual sort implementation has been refactored into two functions in
`sort_utils.rs`: `sort`, and `sort_by`. These can also be used in other
parts of Nushell where values need to be sorted.
- `sort` and `sort-by` used to handle `-i` and `-n` differently.
- `sort -n` would consider all values which can't be coerced into a
string to be equal
- `sort-by -i` and `sort-by -n` would only work if all values were
strings
- In this PR, insensitive sort only affects comparison between strings,
and natural sort only applies to numbers and strings (see below).
- (not a change) Before and after this PR, `sort` and `sort-by` support
sorting mixed types. There was a lot of discussion about potentially
making `sort` and `sort-by` only work on lists of homogeneous types, but
the general consensus was that `sort` should not error just because its
input contains incompatible types.
- In order to try to make working with data containing `null` values
easier, I changed the PartialOrd order to sort `Nothing` values to the
end of a list, regardless of what other types the list contains. Before,
`null` would be sorted before `Binary`, `CellPath`, and `Custom` values.
- (not a change) When sorted, lists of mixed types will contain sorted
values of each type in order, for the most part
- (not a change) For example, `[0x[1] (date now) "a" ("yesterday" | into
datetime) "b" 0x[0]]` will be sorted as `["a", "b", a day ago, now, [0],
[1]]`, where sorted strings appear first, then sorted datetimes, etc.
- (not a change) The exception to this is `Int`s and `Float`s, which
will intermix, `Strings` and `Glob`s, which will intermix, and `None` as
described above. Additionally, natural sort will intermix strings with
ints and floats (see below).
- Natural sort no longer coerce all inputs to strings.
- I did originally make natural only apply to strings, but @fdncred
pointed out that the previous behavior also allowed you to sort numeric
strings with numbers. This seems like a useful feature if we are trying
to support sorting with mixed types, so I settled on coercing only
numbers (int, float). This can be reverted if people don't like it.
- Here is an example of this behavior in action, which is the same
before and after this PR:
```nushell
$ [1 "4" 3 "2"] | sort --natural
╭───┬───╮
│ 0 │ 1 │
│ 1 │ 2 │
│ 2 │ 3 │
│ 3 │ 4 │
╰───┴───╯
```
# User-Facing Changes
## New features
- Replaces the `columns` string parameter of `sort-by` with a cell path
or a closure.
- The cell path parameter works exactly as you would expect
- By default, the `closure` parameter acts as a "key sort"; that is,
each element is transformed by the closure into a sorting key
- With the `--custom` (`-c`) parameter, you can define a comparison
function for completely custom sorting order.
## Behavior changes
<details>
<summary><code>sort -v</code> does not coerce record values to
strings</summary>
This was a bit of a surprising behavior, and is now unified with the
behavior of `sort` and `sort-by`. Here's an example where you can
observe the values being implicitly coerced into strings for sorting, as
they are sorted like strings rather than numbers:
Old behavior:
```nushell
$ {foo: 9 bar: 10} | sort -v
╭─────┬────╮
│ bar │ 10 │
│ foo │ 9 │
╰─────┴────╯
```
New behavior:
```nushell
$ {foo: 9 bar: 10} | sort -v
╭─────┬────╮
│ foo │ 9 │
│ bar │ 10 │
╰─────┴────╯
```
</details>
<details>
<summary>Changed <code>sort-by</code> parameters from
<code>string</code> to <code>cell-path</code> or <code>closure</code>.
Typical interactive usage is the same as before, but if passing a
variable to <code>sort-by</code> it must be a cell path (or closure),
not a string</summary>
Old behavior:
```nushell
$ let sort = "modified"
$ ls | sort-by $sort
╭───┬──────┬──────┬──────┬────────────────╮
│ # │ name │ type │ size │ modified │
├───┼──────┼──────┼──────┼────────────────┤
│ 0 │ foo │ file │ 0 B │ 10 hours ago │
│ 1 │ bar │ file │ 0 B │ 35 seconds ago │
╰───┴──────┴──────┴──────┴────────────────╯
```
New behavior:
```nushell
$ let sort = "modified"
$ ls | sort-by $sort
Error: nu:🐚:type_mismatch
× Type mismatch.
╭─[entry #10:1:14]
1 │ ls | sort-by $sort
· ──┬──
· ╰── Cannot sort using a value which is not a cell path or closure
╰────
$ let sort = $."modified"
$ ls | sort-by $sort
╭───┬──────┬──────┬──────┬───────────────╮
│ # │ name │ type │ size │ modified │
├───┼──────┼──────┼──────┼───────────────┤
│ 0 │ foo │ file │ 0 B │ 10 hours ago │
│ 1 │ bar │ file │ 0 B │ 2 minutes ago │
╰───┴──────┴──────┴──────┴───────────────╯
```
</details>
<details>
<summary>Insensitve and natural sorting behavior reworked</summary>
Previously, the `-i` and `-n` worked differently for `sort` and
`sort-by` (see "Making sort more consistent"). Here are examples of how
these options result in different sorts now:
1. `sort -n`
- Old behavior (types other than numbers, strings, dates, and binary
sorted incorrectly)
```nushell
$ [2sec 1sec] | sort -n
╭───┬──────╮
│ 0 │ 2sec │
│ 1 │ 1sec │
╰───┴──────╯
```
- New behavior
```nushell
$ [2sec 1sec] | sort -n
╭───┬──────╮
│ 0 │ 1sec │
│ 1 │ 2sec │
╰───┴──────╯
```
2. `sort-by -i`
- Old behavior (uppercase words appear before lowercase words as they
would in a typical sort, indicating this is not actually an insensitive
sort)
```nushell
$ ["BAR" "bar" "foo" 2 "FOO" 1] | wrap a | sort-by -i a
╭───┬─────╮
│ # │ a │
├───┼─────┤
│ 0 │ 1 │
│ 1 │ 2 │
│ 2 │ BAR │
│ 3 │ FOO │
│ 4 │ bar │
│ 5 │ foo │
╰───┴─────╯
```
- New behavior (strings are sorted stably, indicating this is an
insensitive sort)
```nushell
$ ["BAR" "bar" "foo" 2 "FOO" 1] | wrap a | sort-by -i a
╭───┬─────╮
│ # │ a │
├───┼─────┤
│ 0 │ 1 │
│ 1 │ 2 │
│ 2 │ BAR │
│ 3 │ bar │
│ 4 │ foo │
│ 5 │ FOO │
╰───┴─────╯
```
3. `sort-by -n`
- Old behavior (natural sort does not work when data contains non-string
values)
```nushell
$ ["10" 8 "9"] | wrap a | sort-by -n a
╭───┬────╮
│ # │ a │
├───┼────┤
│ 0 │ 8 │
│ 1 │ 10 │
│ 2 │ 9 │
╰───┴────╯
```
- New behavior
```nushell
$ ["10" 8 "9"] | wrap a | sort-by -n a
╭───┬────╮
│ # │ a │
├───┼────┤
│ 0 │ 8 │
│ 1 │ 9 │
│ 2 │ 10 │
╰───┴────╯
```
</details>
<details>
<summary>
Sorting a list of non-record values with a non-existent column/path now
errors instead of sorting the values directly (<code>sort</code> should
be used for this, not <code>sort-by</code>)
</summary>
Old behavior:
```nushell
$ [2 1] | sort-by foo
╭───┬───╮
│ 0 │ 1 │
│ 1 │ 2 │
╰───┴───╯
```
New behavior:
```nushell
$ [2 1] | sort-by foo
Error: nu:🐚:incompatible_path_access
× Data cannot be accessed with a cell path
╭─[entry #29:1:17]
1 │ [2 1] | sort-by foo
· ─┬─
· ╰── int doesn't support cell paths
╰────
```
</details>
<details>
<summary><code>sort</code> and <code>sort-by</code> output
<code>List</code> instead of <code>ListStream</code> </summary>
This isn't a meaningful change (unless I misunderstand the purpose of
ListStream), since `sort` and `sort-by` both need to collect in order to
do the sorting anyway, but is user observable.
Old behavior:
```nushell
$ ls | sort | describe -d
╭──────────┬───────────────────╮
│ type │ stream │
│ origin │ nushell │
│ subtype │ {record 3 fields} │
│ metadata │ {record 1 field} │
╰──────────┴───────────────────╯
```
```nushell
$ ls | sort-by name | describe -d
╭──────────┬───────────────────╮
│ type │ stream │
│ origin │ nushell │
│ subtype │ {record 3 fields} │
│ metadata │ {record 1 field} │
╰──────────┴───────────────────╯
```
New behavior:
```nushell
ls | sort | describe -d
╭────────┬─────────────────╮
│ type │ list │
│ length │ 22 │
│ values │ [table 22 rows] │
╰────────┴─────────────────╯
```
```nushell
$ ls | sort-by name | describe -d
╭────────┬─────────────────╮
│ type │ list │
│ length │ 22 │
│ values │ [table 22 rows] │
╰────────┴─────────────────╯
```
</details>
- `sort` now errors when nothing is piped in (`sort-by` already did
this)
# Tests + Formatting
I added lots of unit tests on the new sort implementation to enforce new
sort behaviors and prevent regressions.
# After Submitting
See [docs PR](https://github.com/nushell/nushell.github.io/pull/1568),
which is ~2/3 finished.
---------
Co-authored-by: NotTheDr01ds <32344964+NotTheDr01ds@users.noreply.github.com>
Co-authored-by: Ian Manske <ian.manske@pm.me>
# Description
In this PR I replaced most of the raw usize IDs with
[newtypes](https://doc.rust-lang.org/rust-by-example/generics/new_types.html).
Some other IDs already started using new types and in this PR I did not
want to touch them. To make the implementation less repetitive, I made
use of a generic `Id<T>` with marker structs. If this lands I would try
to move make other IDs also in this pattern.
Also at some places I needed to use `cast`, I'm not sure if the type was
incorrect and therefore casting not needed or if actually different ID
types intermingle sometimes.
# User-Facing Changes
Probably few, if you got a `DeclId` via a function and placed it later
again it will still work.
# Description
Cleans up and refactors the config code using the `IntoValue` macro.
Shoutout to @cptpiepmatz for making the macro!
# User-Facing Changes
Should be none.
# After Submitting
Somehow refactor the reverse transformation.
# Description
This PR allows the helper attribute `nu_value(rename = "...")` to be
used on struct fields and enum variants. This allows renaming keys and
variants just like [`#[serde(rename =
"name")]`](https://serde.rs/field-attrs.html#rename). This has no
singular variants for `IntoValue` or `FromValue`, both need to use the
same (but I think this shouldn't be an issue for now).
# User-Facing Changes
Users of the derive macros `IntoValue` and `FromValue` may now use
`#[nu_value(rename = "...")]` to rename single fields, but no already
existing code will break.
# Description
Implements `IntoValue` for `&str` and `DateTime` as well as other
nushell types like `Record` and `Closure`. Also allows `HashMap`s with
keys besides `String` to implement `IntoValue`.
<!--
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.
-->
@sholderbach mentioned that I introduced `convert_case` as a dependency
while we already had `heck` for case conversion. So in this PR replaced
the use `convert_case` with `heck`. Mostly I rebuilt the `convert_case`
API with `heck` to work with it as I like the API of `convert_case` more
than `heck`.
# User-Facing Changes
<!-- List of all changes that impact the user experience here. This
helps us keep track of breaking changes. -->
Nothing changed, the use of `convert_case` wasn't exposed anywhere and
all case conversions are still available.
# 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 toolkit.nu; toolkit test stdlib"` 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
> ```
-->
No new tests required but my tests in `test_derive` captured some errors
I made while developing this change, (hurray, tests work 🎉)
- 🟢 `toolkit fmt`
- 🟢 `toolkit clippy`
- 🟢 `toolkit test`
- 🟢 `toolkit test stdlib`
# 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
Using derived `IntoValue` and `FromValue` implementations on structs
with named fields currently produce `Value::Record`s where each key is
the key of the Rust struct. For records like the `$nu` constant, that
won't work as this record uses `kebab-case` for it's keys. To accomodate
this, I upgraded the `#[nu_value(rename_all = "...")]` helper attribute
to also work on structs with named fields which will rename the keys via
the same case conversion as the enums already have.
# User-Facing Changes
Users of these macros may choose different key styles for their in
`Value` representation.
# Tests + Formatting
I added the same test suite as enums already have and updated the traits
documentation with more examples that also pass the doc test.
# After Submitting
I played around with the `$nu` constant but got stuck at the point that
these keys are kebab-cased, with this, I can play around more with it.
<!--
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.
-->
In this PR I expanded the helper attribute `#[nu_value]` on
`#[derive(FromValue)]`. It now allows the usage of `#[nu_value(type_name
= "...")]` to set a type name for the `FromValue::expected_type`
implementation. Currently it only uses the default implementation but
I'd like to change that without having to manually implement the entire
trait on my own.
# User-Facing Changes
<!-- List of all changes that impact the user experience here. This
helps us keep track of breaking changes. -->
Users that derive `FromValue` may now change the name of the expected
type.
# 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 toolkit.nu; toolkit test stdlib"` 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
> ```
-->
I added some tests that check if this feature work and updated the
documentation about the derive macro.
- 🟢 `toolkit fmt`
- 🟢 `toolkit clippy`
- 🟢 `toolkit test`
- 🟢 `toolkit test stdlib`
# 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.
-->
<!--
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.
-->
I was working with byte collections like `Vec<u8>` and
[`bytes::Bytes`](https://docs.rs/bytes/1.7.1/bytes/struct.Bytes.html),
both are currently not possible to be used directly in a struct that
derives `IntoValue` and `FromValue` at the same time. The `Vec<u8>` will
convert itself into a `Value::List` but expects a `Value::String` or
`Value::Binary` to load from. I now also implemented that it can load
from `Value::List` just like the other `Vec<uX>` versions. For further
working with byte collections the type `bytes::Bytes` is wildly used,
therefore I added a implementation for it. `bytes` is already part of
the dependency graph as many crates (more than 5000 to crates.io) use
it.
# User-Facing Changes
<!-- List of all changes that impact the user experience here. This
helps us keep track of breaking changes. -->
User of `nu-protocol` as library, e.g. plugin developers, can now use
byte collections more easily in their data structures and derive
`IntoValue` and `FromValue` for it.
# 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 toolkit.nu; toolkit test stdlib"` 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
> ```
-->
I added a few tests that check that these byte collections are correctly
translated in and from `Value`. They live in `test_derive.rs` as part of
the `ByteContainer` and I also explicitely tested that `FromValue` for
`Vec<u8>` works as expected.
- 🟢 `toolkit fmt`
- 🟢 `toolkit clippy`
- 🟢 `toolkit test`
- 🟢 `toolkit test stdlib`
# 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.
-->
Maybe it should be explored if `Value::Binary` should use `bytes::Bytes`
instead of `Vec<u8>`.
# Description
Fixes#11267
Shifting by a `shift >= num_bits` is undefined in the underlying
operation. Previously we also had an overflow on negative shifts for the
operators `bit-shl` and `bit-shr`
Furthermore I found a severe bug in the implementation of shifting of
`binary` data with the commands `bits shl` and `bits shr`, this
categorically produced incorrect results with shifts that were not
`shift % 4 == 0`. `bits shr` also was able to produce outputs with
different size to the input if the shift was exceeding the length of the
input data by more than a byte.
# User-Facing Changes
It is now an error trying to shift by more than the available bits with:
- `bit-shl` operator
- `bit-shr` operator
- command `bits shl`
- command `bits shr`
# Tests + Formatting
Added testing for all relevant cases
In some `if let`s we ran the `SharedCow::to_mut` for the test and to get
access to a mutable reference in the happy path. Internally
`Arc::into_mut` has to read atomics and if necessary clone.
For else branches, where we still want to modify the record we
previously called this again (not just in rust, confirmed in the asm).
This would have introduced a `call` instruction and its cost (even if it
would be guaranteed to take the short path in `Arc::into_mut`).
Lifting it get's rid of this.
# Description
This PR introduces a new `Signals` struct to replace our adhoc passing
around of `ctrlc: Option<Arc<AtomicBool>>`. Doing so has a few benefits:
- We can better enforce when/where resetting or triggering an interrupt
is allowed.
- Consolidates `nu_utils::ctrl_c::was_pressed` and other ad-hoc
re-implementations into a single place: `Signals::check`.
- This allows us to add other types of signals later if we want. E.g.,
exiting or suspension.
- Similarly, we can more easily change the underlying implementation if
we need to in the future.
- Places that used to have a `ctrlc` of `None` now use
`Signals::empty()`, so we can double check these usages for correctness
in the future.
# Description
In #13031 I added the derive macros for `FromValue` and `IntoValue`. In
that implementation, in particular for structs with named fields, it was
not possible to omit fields while loading them from a value, when the
field is an `Option`. This PR adds extra handling for this behavior, so
if a field is an `Option` and that field is missing in the `Value`, then
the field becomes `None`. This behavior is also tested in
`nu_protocol::value::test_derive::missing_options`.
# User-Facing Changes
When using structs for options or similar, users can now just emit
fields in the record and the derive `from_value` method will be able to
understand this, if the struct has an `Option` type for that field.
# Tests + Formatting
- 🟢 `toolkit fmt`
- 🟢 `toolkit clippy`
- 🟢 `toolkit test`
- 🟢 `toolkit test stdlib`
# After Submitting
A showcase for this feature would be great, I tried to use the current
derive macro in a plugin of mine for a config but without this addition,
they are annoying to use. So, when this is done, I would add an example
for such plugin configs that may be loaded via `FromValue`.
# Description
After discussing with @sholderbach the cumbersome usage of
`nu_protocol::Value` in Rust, I created a derive macro to simplify it.
I’ve added a new crate called `nu-derive-value`, which includes two
macros, `IntoValue` and `FromValue`. These are re-exported in
`nu-protocol` and should be encouraged to be used via that re-export.
The macros ensure that all types can easily convert from and into
`Value`. For example, as a plugin author, you can define your plugin
configuration using a Rust struct and easily convert it using
`FromValue`. This makes plugin configuration less of a hassle.
I introduced the `IntoValue` trait for a standardized approach to
converting values into `Value` (and a fallible variant `TryIntoValue`).
This trait could potentially replace existing `into_value` methods.
Along with this, I've implemented `FromValue` for several standard types
and refined other implementations to use blanket implementations where
applicable.
I made these design choices with input from @devyn.
There are more improvements possible, but this is a solid start and the
PR is already quite substantial.
# User-Facing Changes
For `nu-protocol` users, these changes simplify the handling of
`Value`s. There are no changes for end-users of nushell itself.
# Tests + Formatting
Documenting the macros itself is not really possible, as they cannot
really reference any other types since they are the root of the
dependency graph. The standard library has the same problem
([std::Debug](https://doc.rust-lang.org/stable/std/fmt/derive.Debug.html)).
However I documented the `FromValue` and `IntoValue` traits completely.
For testing, I made of use `proc-macro2` in the derive macro code. This
would allow testing the generated source code. Instead I just tested
that the derived functionality is correct. This is done in
`nu_protocol::value::test_derive`, as a consumer of `nu-derive-value`
needs to do the testing of the macro usage. I think that these tests
should provide a stable baseline so that users can be sure that the impl
works.
# After Submitting
With these macros available, we can probably use them in some examples
for plugins to showcase the use of them.
# Description
On 64-bit platforms the current size of `Value` is 56 bytes. The
limiting variants were `Closure` and `Range`. Boxing the two reduces the
size of Value to 48 bytes. This is the minimal size possible with our
current 16-byte `Span` and any 24-byte `Vec` container which we use in
several variants. (Note the extra full 8-bytes necessary for the
discriminant or other smaller values due to the 8-byte alignment of
`usize`)
This is leads to a size reduction of ~15% for `Value` and should overall
be beneficial as both `Range` and `Closure` are rarely used compared to
the primitive types or even our general container types.
# User-Facing Changes
Less memory used, potential runtime benefits.
(Too late in the evening to run the benchmarks myself right now)
# Description
Removes lazy records from the language, following from the reasons
outlined in #12622. Namely, this should make semantics more clear and
will eliminate concerns regarding maintainability.
# User-Facing Changes
- Breaking change: `lazy make` is removed.
- Breaking change: `describe --collect-lazyrecords` flag is removed.
- `sys` and `debug info` now return regular records.
# After Submitting
- Update nushell book if necessary.
- Explore new `sys` and `debug info` APIs to prevent them from taking
too long (e.g., subcommands or taking an optional column/cell-path
argument).
# Description
Continuing from #12568, this PR further reduces the size of `Expr` from
64 to 40 bytes. It also reduces `Expression` from 128 to 96 bytes and
`Type` from 32 to 24 bytes.
This was accomplished by:
- for `Expr` with multiple fields (e.g., `Expr::Thing(A, B, C)`),
merging the fields into new AST struct types and then boxing this struct
(e.g. `Expr::Thing(Box<ABC>)`).
- replacing `Vec<T>` with `Box<[T]>` in multiple places. `Expr`s and
`Expression`s should rarely be mutated, if at all, so this optimization
makes sense.
By reducing the size of these types, I didn't notice a large performance
improvement (at least compared to #12568). But this PR does reduce the
memory usage of nushell. My config is somewhat light so I only noticed a
difference of 1.4MiB (38.9MiB vs 37.5MiB).
---------
Co-authored-by: Stefan Holderbach <sholderbach@users.noreply.github.com>
# Description
`Value` describes the types of first-class values that users and scripts
can create, manipulate, pass around, and store. However, `Block`s are
not first-class values in the language, so this PR removes it from
`Value`. This removes some unnecessary code, and this change should be
invisible to the user except for the change to `scope modules` described
below.
# User-Facing Changes
Breaking change: the output of `scope modules` was changed so that
`env_block` is now `has_env_block` which is a boolean value instead of a
`Block`.
# After Submitting
Update the language guide possibly.
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
Currently, `Range` is a struct with a `from`, `to`, and `incr` field,
which are all type `Value`. This PR changes `Range` to be an enum over
`IntRange` and `FloatRange` for better type safety / stronger compile
time guarantees.
Fixes: #11778Fixes: #11777Fixes: #11776Fixes: #11775Fixes: #11774Fixes: #11773Fixes: #11769.
# User-Facing Changes
Hopefully none, besides bug fixes.
Although, the `serde` representation might have changed.
# 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 keeps plugin custom values that have requested drop notification
around during the lifetime of a plugin call / stream by sending them to
a channel that gets persisted during the lifetime of the call.
Before this change, it was very likely that the drop notification would
be sent before the plugin ever had a chance to handle the value it
received.
Tests have been added to make sure this works - see the `custom_values`
plugin.
cc @ayax79
# User-Facing Changes
This is basically just a bugfix, just a slightly big one.
However, I did add an `as_mut_any()` function for custom values, to
avoid having to clone them. This is a breaking change.
# 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
This clone is not necessary and tanks the performance of deep nested
access.
As soon as we found the value, we know we discard the old value, so can
`std::mem::take` the inner (`impl Default for Value` to the rescue)
We may be able to further optimize this but not having to clone the
value is vital.
# Description
The second `Value` is redundant and will consume five extra bytes on
each transmission of a custom value to/from a plugin.
# User-Facing Changes
This is a breaking change to the plugin protocol.
The [example in the protocol
reference](https://www.nushell.sh/contributor-book/plugin_protocol_reference.html#value)
becomes
```json
{
"Custom": {
"val": {
"type": "PluginCustomValue",
"name": "database",
"data": [36, 190, 127, 40, 12, 3, 46, 83],
"notify_on_drop": true
},
"span": {
"start": 320,
"end": 340
}
}
}
```
instead of
```json
{
"CustomValue": {
...
}
}
```
# After Submitting
Update plugin protocol reference
# Description
When implementing a `Command`, one must also import all the types
present in the function signatures for `Command`. This makes it so that
we often import the same set of types in each command implementation
file. E.g., something like this:
```rust
use nu_protocol::ast::Call;
use nu_protocol::engine::{Command, EngineState, Stack};
use nu_protocol::{
record, Category, Example, IntoInterruptiblePipelineData, IntoPipelineData, PipelineData,
ShellError, Signature, Span, Type, Value,
};
```
This PR adds the `nu_engine::command_prelude` module which contains the
necessary and commonly used types to implement a `Command`:
```rust
// command_prelude.rs
pub use crate::CallExt;
pub use nu_protocol::{
ast::{Call, CellPath},
engine::{Command, EngineState, Stack},
record, Category, Example, IntoInterruptiblePipelineData, IntoPipelineData, IntoSpanned,
PipelineData, Record, ShellError, Signature, Span, Spanned, SyntaxShape, Type, Value,
};
```
This should reduce the boilerplate needed to implement a command and
also gives us a place to track the breadth of the `Command` API. I tried
to be conservative with what went into the prelude modules, since it
might be hard/annoying to remove items from the prelude in the future.
Let me know if something should be included or excluded.
<!--
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.
-->
Boxes `Record` inside `Value` to reduce memory usage, `Value` goes from
`72` -> `56` bytes after this change.
# 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
Fixes: #11887Fixes: #11626
This pr unify the tilde expand behavior over several filesystem relative
commands. It follows the same rule with glob expansion:
| command | result |
| ----------- | ------ |
| ls ~/aaa | expand tilde
| ls "~/aaa" | don't expand tilde
| let f = "~/aaa"; ls $f | don't expand tilde, if you want to: use `ls
($f \| path expand)`
| let f: glob = "~/aaa"; ls $f | expand tilde, they don't expand on
`mkdir`, `touch` comamnd.
Actually I'm not sure for 4th item, currently it's expanding is just
because it followes the same rule with glob expansion.
### About the change
It changes `expand_path_with` to accept a new argument called
`expand_tilde`, if it's true, expand it, if not, just keep it as `~`
itself.
# User-Facing Changes
After this change, `ls "~/aaa"` won't expand tilde.
# Tests + Formatting
Done
