# Description
Fixes: #7085
Also closes: #7526
# User-Facing Changes
After this change, we need to use `-c` flag like this:
```nushell
[[a, b, c]; [1, 2, 3]] | rename -c { a: ham }
```
But we can rename many columns easily, here is another example:
```nushell
[[a, b, c]; [1, 2, 3]] | rename -c { a: ham, b: ham2 }
```
<!--
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
This fixes the default prompt on Windows to use the correct path
direction
# 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.
-->
Align the clippy checks done by `toolkit check pr` with the ones done by
github CI when you post a PR.
Motivated in no small part by my frustration at having the third PR in
one day fail due to a lint that `toolkit check pr` didn't find!
# Description
Change definition of `toolkit clippy`
* always run checks on all crates in the workspace
* always run checks on tests as well as non-test code
* don't check for `unwrap_used`. There are many (legit) uses of
`.unwrap()` in tests,
*
# User-Facing Changes
Changes to `toolkit clippy`:
* remove `--workspace` switch
* `--features` is no longer mutually exclusive with `--workspace`
# Tests + Formatting
n/a
# 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.
-->
---------
Co-authored-by: amtoine <stevan.antoine@gmail.com>
# Description
The description `Custom` doesn't really reflect meaning in the set of
`SyntaxShape`. Makes it a bit more verbose but explicit
# User-Facing Changes
Only hypothetically breaking as plugins can not effectively use a
requirement on `SyntaxShape::Custom`.
# Tests + Formatting
(-)
related to
- https://github.com/nushell/nushell/pull/10532
# Description
i was reviewing https://github.com/nushell/nushell/pull/10532 and
thought
> wait a minute, this line is huge and it's basically impossible to
review properly...
i had to grab the diff and throw some Nushell magic at it to see that it
was valid 😱
in this PR, i just split the loooooong string on the `:`, put that in a
list, join with `.join(":")` and borrow that to get a `str` 👌
# User-Facing Changes
# Tests + Formatting
# After Submitting
# Description
Closes#10441
Uses `String::to_lowercase()` when the file's extension `ext` is parsed
to allow `from_decl(format!("from {ext}"))` to return the desired output
regardless of extension case.
It doesn't work with sqlite files since those are handled earlier in the
parsing but I think is good- since there's no standard file extension
used by sqlite so a user will likely want case sensitivity in that case.
This also has the (possibly undesired) effect of making `open`
completely case insensitive, e.g. `open foo.JSON` will work on a file
named `foo.json` and vice versa. This is good on Windows as it treats
`foo.json` and `foo.JSON` as the same file, but may not be the desired
behaviour on Unix.
If this behaviour is undesired I assume it would be fixed with a
`#[cfg(not(unix))]` attribute on the `to_lowercase()` operation but that
produces slightly "uglier" code that I didn't wish to submit unless
necessary.
old behaviour:
new behaviour:
# User-Facing Changes
`open` will now present a table when `open`-ing files with captitalized
extensions rather than the file's raw data
# Tests + Formatting
new test: `parses_file_with_uppercase_extension` which tests the desired
behaviour
---------
Co-authored-by: Stefan Holderbach <sholderbach@users.noreply.github.com>
This pr closes#10521.
# Description
The default prompt by nushell will replace `$nu.home-path` with `~`.
E.g. for a user named `user`, `/home/user` would become `~`. This also
works with sub paths, e.g. `/home/user/docs` would become `~/docs`.
The issue is that this replacement was a tad overzealous. A path like
`/home/user-with-suffix` would become `~-with-suffix`. This PR fixes the
issue by updating the home path detection logic.
# User-Facing Changes
The bugged behavior no longer occurs.
# Tests + Formatting
* `cargo` checks were not performed as this does not touch rust.
* The updated logic was tested against
[elvish](https://github.com/elves/elvish)'s path replacement logic, for
~10,000 randomly selected folders on a linux server. All paths were
processed the same.
# Description
This merges @horasal 's changes from #10246 and #10269Closes#10205Closes#8714
Fixes the bug that editor paths with spaces are unusable
Closes#10210Closes#10269
# User-Facing Changes
You can now either pass a string with the name of the executable or a
list with the executable and any flags to
`$env.config.buffer_editor`/`$env.EDITOR`/`$env.VISUAL`
Both the external buffer editor of reedline (by default bound to
`Ctrl-o`) and the commands `config nu` and `config env` will respect
those variables in the following order:
1. `$env.config.buffer_editor`
2. `$env.EDITOR`
3. `$env.VISUAL`
Example:
```
$env.EDITOR = "nvim" # The system-wide EDITOR is neovim
$env.config.buffer_editor = ["vim" "-p2"] # Force vim to open two tabs (not particularly useful)
$env.config.buffer_editor = null # Unset `buffer_editor` -> Uses `$env.EDITOR` ergo nvim
```
# Tests + Formatting
None
---------
Co-authored-by: Horasal <1991933+horasal@users.noreply.github.com>
# Description
Support keyboard enhancement protocol as implemented by Kitty console,
hence Kitty protocol.
This PR enables Nushell to use keybinding that is not available before,
such as Ctrl+i (that alias to Tab) or Ctrl+e (that alias to Esc, likely
I mistaken). After this PR merged and you set `use_kitty_protocol`
enabled, if your console app support Kitty protocol (WezTerm, Kitty,
etc.) you will be able to set more fine-grained keybinding.
For Colemak users, this feature is a blessing, because some Ctrl+[hjkl]
that previously unmap-able to Ctlr+[hnei] now it is.
# User-Facing Changes
This adds `use_kitty_protocol` config which defaults to false. When set
to `true`, it enables kitty protocol on the line editor when supported,
or else it warns.
---------
Co-authored-by: Stefan Holderbach <sholderbach@users.noreply.github.com>
Fixes#10503
Also improves link to metacharacter help;
# Description
`glob` code was using pattern as provided by user. If that had leading
`..\`, `wax::Glob` is documented to treat them as literal chars to be
matched.
Fix is to use `wax::Glob.partition()` to split such invariant prefixes
off the pattern and tack them onto the working directory computed
separately.
Before
```
> ls ..
╭───┬───────┬──────┬──────┬───────────────╮
│ # │ name │ type │ size │ modified │
├───┼───────┼──────┼──────┼───────────────┤
│ 0 │ ../r1 │ dir │ 7 B │ 3 hours ago │
│ 1 │ ../r2 │ dir │ 3 B │ a day ago │
│ 2 │ ../r3 │ dir │ 13 B │ 4 minutes ago │
╰───┴───────┴──────┴──────┴───────────────╯
> glob ../r*
╭────────────╮
│ empty list │
╰────────────╯
```
After
```
> glob ../r*
╭───┬──────────────────────────────╮
│ 0 │ /home/bobhy/src/rust/work/r2 │
│ 1 │ /home/bobhy/src/rust/work/r1 │
│ 2 │ /home/bobhy/src/rust/work/r3 │
╰───┴──────────────────────────────╯
```
# User-Facing Changes
None
# Tests + Formatting
- 🟢 `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.
-->
---------
Co-authored-by: Darren Schroeder <343840+fdncred@users.noreply.github.com>
More incremental `explore` improvements!
This PR removes the `show_cursor` config from the `explore` command, in
favour of always using the background colour to highlight the selected
cell. I believe this is a better default and I'd like to remove the
`show_cursor` functionality entirely as part of the effort to simplify
`explore`.
The style for selected cells is still configurable. I went with light
blue for the default background colour, it looks OK to me.
## Before:
## After:
# Description
Those `SyntaxShape`s can not coerce to `Value`s or `Type`s that can be
used by the user in an argument or input-output-type position.
Supporting them doesn't make sense.
# User-Facing Changes
Removal of useless "types" in argument type or input/output type
positions
# Tests + Formatting
No adjustment necessary
close#10396
# Description
Change LS_COLORS variable to bring the highlighting for .fb2 files in
line with other types of text documents
### Before
### After
# Description
Fixes#2047 but for the `doas` command the same way as in #8094
# User-Facing Changes
No breaking changes. If people not using `doas`, no difference at all.
# Tests
I have not added any tests since its using same logic as for "sudo". I
guess if something would go wrong in this part, sudo tests will cover
it?
# Additional context
As a nushell user I could not find a way to implement custom completion
for a "sudo like command". Since I can see `sudo` being hardcoded in
sources, this is what I propose.
~~Also I have almost zero knowledge of rust and this is definitely not
the clean way yet~~
---------
Co-authored-by: Stefan Holderbach <sholderbach@users.noreply.github.com>
# Description
We don't use this shape during parsing and never reference it in command
signatures. Thus it should be removed.
# User-Facing Changes
None functional.
Plugin authors that used it would never be provided with data that
specifically matched `SyntaxShape::Variable`
Builds using it will now fail.
# Tests + Formatting
NA
# Description
Consistently use `int` for types and commands
h/t @1kinoti
Work for #10332
# User-Facing Changes
Deprecate `random integer` in the next release
New command `random int`
# Tests + Formatting
(-)
Closes https://github.com/nushell/nushell/issues/8847
# Description
If the `HTTP_PROXY` variable is found, use its value to setup ureq
proxy. I haven't implemented `NO_PROXY` at the moment.
# User-Facing Changes
No breaking change for the user, the network commands simply use an
environment variable.
# Tests + Formatting
The existing tests seem to run fine, although I can't think of a new
test to add.
related to
- https://github.com/nushell/nushell/pull/9973
- https://github.com/nushell/nushell/pull/9918
thanks to @jntrnr and their super useful tips on this PR, i learned
about the parser + evaluation, so 🙏
# Description
because we already have `null` as the value of the type `nothing` and as
a followup to the two other attempts of mine, i propose to remove the
redundant `$nothing` built-in variable 😋
this PR is the first step, deprecating `$nothing`.
a followup PR will remove it altogether and wait for 0.87 👍⚙️ **details**: a new `NOTHING_VARIABLE_ID = 3` has been added,
parsing `$nothing` will create it, finally a `Value::Nothing` will be
produced and a warning will be reported.
this PR already fixes the `toolkit.nu` module so that it does not throw
a bunch of warnings each time 👌
# User-Facing Changes
`$nothing` is now deprecated and will be removed in 0.87
```nushell
> $nothing
Error: × Deprecated variable
╭─[entry #1:1:1]
1 │ $nothing
· ────┬───
· ╰── `$nothing` is deprecated and will be removed in 0.87.
╰────
help: Use `null` instead
```
# Tests + Formatting
tests have been updated, especially
- `nothing_fails_string`
- `nothing_fails_int`
which use a variable called `nil` now to make sure `nothing` does not
support cell paths 👍
# After Submitting
classic deprecation mention 👍
Fixes: #10476
After the change, the error message will be something like this:
```nushell
❯ not null
Error: nu:🐚:type_mismatch
× Type mismatch.
╭─[entry #11:1:1]
1 │ not null
· ──┬─
· ╰── expected bool, found nothing
╰────
```
should close https://github.com/nushell/nushell/issues/10406
# Description
when writing a script, with variables you try to `ls` or `open`, you
will get a "directory not found" error but the variable won't be
expanded and you won't be able to see which one of the variable was the
issue...
this PR adds this information to the error.
# User-Facing Changes
let's define a variable
```nushell
let does_not_exist = "i_do_not_exist_in_the_current_directory"
```
### before
```nushell
> open $does_not_exist
Error: nu:🐚:directory_not_found
× Directory not found
╭─[entry #7:1:1]
1 │ open $does_not_exist
· ───────┬───────
· ╰── directory not found
╰────
```
```nushell
> ls $does_not_exist
Error: nu:🐚:directory_not_found
× Directory not found
╭─[entry #8:1:1]
1 │ ls $does_not_exist
· ───────┬───────
· ╰── directory not found
╰────
```
### after
```nushell
> open $does_not_exist
Error: nu:🐚:directory_not_found
× Directory not found
╭─[entry #3:1:1]
1 │ open $does_not_exist
· ───────┬───────
· ╰── directory not found
╰────
help: /home/amtoine/documents/repos/github.com/amtoine/nushell/i_do_not_exist_in_the_current_directory does not exist
```
```nushell
> ls $does_not_exist
Error: nu:🐚:directory_not_found
× Directory not found
╭─[entry #4:1:1]
1 │ ls $does_not_exist
· ───────┬───────
· ╰── directory not found
╰────
help: /home/amtoine/documents/repos/github.com/amtoine/nushell/i_do_not_exist_in_the_current_directory does not exist
```
# Tests + Formatting
shouldn't harm anything 🤞
# After Submitting
<!--
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.
-->
Closes#5436
When I opened this issue more than a year ago, I mainly wanted the
following capacity: easily access the full env and have the hability to
update it when a new version of `nushell` comes out.
With this PR I can now do the following:
```nu
source-env ~/.config/nushell/defaults/env.nu
source ~/.config/nushell/defaults/config.nu
# Update nushell default config & env file (run this after a version update)
def update-defaults [] {
config env --default | save -f ~/.config/nushell/defaults/env.nu
config nu --default | save -f ~/.config/nushell/defaults/config.nu
}
```
Which is more than enough for me. Along with `nushell` respecting the
XDG spec on macOS (`dirs-next` should be banned for CLI tools on macOS),
this should be one of the last hurdle before fully switching for me!
# User-Facing Changes
<!-- List of all changes that impact the user experience here. This
helps us keep track of breaking changes. -->
Two new switches to existing commands:
```nu
config env --default # Print the default env embedded at compile time in the binary
config nu --default # Print the default config embedded at compile time in the binary
```
# 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
> ```
-->
- Added a test for the output of `config env --default`
- Added a test for the output of `config nu --default`
# 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.
-->
Are the docs for commands generated automatically or do I need to make a
PR there too ? It's no problem if so, just point me at instructions if
there are any :)
# Description
Magenta wasn't being interpreted correctly. note that `bg:
magenta_reverse attr: b` showed up as white. This was because it was
missing from the lookup and it was defaulting to white.
fixes#10490
### Before
### After
# 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.
-->
### Update winget submission workflow include only default msi files:
---
The new release workflow after
https://github.com/nushell/nushell/pull/10457 will add full featured
packages like: `nu-*-x86_64-windows-msvc-full.msi` and
`nu-*-aarch64-windows-msvc-full.msi` (A release preview could be found
here: https://github.com/nushell/nightly/releases/tag/nightly-6a2fd91 )
and this change will make sure only the default packages will be
included just the same as before to make sure that nothing will be
broken.
BTW: Due to lots of changes made recently about the release workflow, I
will stand-by for the next release.
# Description
Fix type checking in arguments default values not adhering to subtyping
rules
Currently following examples produce a parse error:
```nu
def test [ --qwe: record<a: int> = {a: 1 b: 1} ] { }
def test [ --qwe: list<any> = [ 1 2 3 ] ] { }
```
despite types matching. Type equality check is replaced with subtyping
check and everything parses fine:
# User-Facing Changes
Default values of flag arguments type checking behavior is in line with
`let` statements
# Description
This PR fixes#9702 on the side of parse. I.e. input/output types in
signature and type annotations in `let` now should correctly parse with
type annotations that contain commas and spaces:
# User-Facing Changes
Return values and let type annotations now can contain stuff like
`table<a: int b: record<c: string d: datetime>>` e.t.c
fixes#10455
@KAAtheWiseGit, I'm sorry, I didn't mean to block your first PR #10461,
didn't see you had submitted it till I got around to submitting this. If
you want to incoporate useful ideas from this PR into yours, I do not
mind deferring to you.
# Description
Changes made in `datetime-diff`:
* Initialize millisecond and microsecond fields in `$current`, to fix
the error when borrow needs to refer to them.
* Fix `borrow_nanoseconds` to borrow from seconds, not from (unused)
microseconds.
* Added error check to insist that first argument is >= second argument.
`datetime-diff` doesn't represent negative durations correctly (it tries
to borrow out of the year, resulting in negative year and positive all
other fields). We don't currently have a use case requiring negative
durations.
* Add comments so help is a bit clearer (I was surprised that the first
argument, named `$from` was actually supposed to be the *later*
datetime. The order of arguments is reasonable (reminiscent of <later>
<minus> <earlier>), so I just changed the param name to match its
purpose.
Changes made in `pretty-print-duration`:
* changed type of argument from `duration` to `record`. (it's not clear
why Nu was not complaining about this!)
* changed test for skipping a clause from `> 0` to `!= 0`. Even though
`datetime-diff` won't present a negative field in the record, user might
call `pretty-print-duration` with one, might as well handle it. (but I
think `hour:-2` will be rendered as `-2hr`, not `-2hrs`...).
* added help and an example.
# User-Facing Changes
none requiring code changes.
# Tests + Formatting
- 🟢 `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.
-->
related to
- https://github.com/nushell/nushell/pull/10456
# Description
this PR will fix the public API of the standard library by removing the
type annotations from public boolean switches.
1. the signature before
```nushell
clip [--silent: bool, --no-notify: bool, --no-strip: bool, --expand (-e): bool, --codepage (-c): int]
```
2. the signature after
```nushell
clip [--silent, --no-notify, --no-strip, --expand (-e), --codepage (-c): int]
```
# User-Facing Changes
### before
```nushell
> "foo" | clip
Error: nu:🐚:cant_convert
× Can't convert to bool.
╭─[NU_STDLIB_VIRTUAL_DIR/std/mod.nu:148:1]
148 │ $in
149 │ | if $expand { table --expand } else { table }
· ───┬───
· ╰── can't convert nothing to bool
150 │ | into string
╰────
```
### after
```nushell
> "foo" | clip
foo
saved to clipboard
```
# Tests + Formatting
# After Submitting
# Description
Fixes: #10450
This pr differentiating between `--x: bool` and `--x`
Here are examples which demostrate difference between them:
```nushell
def a [--x: bool] { $x };
a --x # not allowed, you need to parse a value to the flag.
a # it's allowed, and the value of `$x` is false, which behaves the same to `def a [--x] { $x }; a`
```
For boolean flag with default value, it works a little bit different to
#10450 mentioned:
```nushell
def foo [--option: bool = false] { $option }
foo # output false
foo --option # not allowed, you need to parse a value to the flag.
foo --option true # output true
```
# User-Facing Changes
After the pr, the following code is not allowed:
```nushell
def a [--x: bool] { $x }; a --x
```
Instead, you have to pass a value to flag `--x` like `a --x false`. But
bare flag works in the same way as before.
## Update: one more breaking change to help on #7260
```
def foo [--option: bool] { $option == null }
foo
```
After the pr, if we don't use a boolean flag, the value will be `null`
instead of `true`. Because here `--option: bool` is treated as a flag
rather than a switch
---------
Co-authored-by: amtoine <stevan.antoine@gmail.com>
# Description
Add full releases of Nu binaries along with the standard releases,
close: https://github.com/nushell/nushell/issues/10322
A full release means:
1. Build with `--features=dataframe,extra` flag
2. Contains `-full` in the package name
### **A test release could be found in the nushell/nightly Repo:
https://github.com/nushell/nightly/releases/tag/v0.85.1**
The action running log:
https://github.com/nushell/nightly/actions/runs/6260611553
# User-Facing Changes
Will attach the following release packages along with the official ones
as before:
1. nu-*-aarch64-darwin-full.tar.gz
2. nu-*-aarch64-linux-gnu-full.tar.gz
3. nu-*-aarch64-windows-msvc-full.msi
4. nu-*-aarch64-windows-msvc-full.zip
5. nu-*-x86_64-darwin-full.tar.gz
6. nu-*-x86_64-linux-gnu-full.tar.gz
7. nu-*-x86_64-linux-musl-full.tar.gz
8. nu-*-x86_64-windows-msvc-full.msi
9. nu-*-x86_64-windows-msvc-full.zip
## Description
This PR uses environment variables to enable and set a transient prompt,
which lets you draw a different prompt once you've entered a command and
you've moved on to the next line. This is useful if you have a fancy
two-line prompt with a bunch of info about time and git status that you
don't really need in your scrollback buffer.
Here's a screenshot. You can see how my usual prompt has two lines and
would take up a lot more space if every past command also used the full
prompt, but reducing past prompts to `🚀` or `>` makes it take up less
space.
I added the following lines to my `env.nu` to get that rocket as the
prompt initially:
```nu
$env.TRANSIENT_PROMPT_COMMAND = {|| "" }
$env.TRANSIENT_PROMPT_INDICATOR = {|| open --raw "~/.prompt-indicator" }
$env.TRANSIENT_PROMPT_INDICATOR_VI_INSERT = $env.TRANSIENT_PROMPT_INDICATOR
```
## User-Facing Changes
If you want to change a segment of the prompt, set the corresponding
`TRANSIENT_PROMPT_*` variable.
<!-- 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.
-->
## Problems/Things to Consider:
- The transient prompt clones the `Stack` at the very beginning of the
session and keeps that around. I'm not sure if that could cause
problems, but if so, it could probably take an `Arc<State>` instead.
- This isn't truly a problem, but now there's even more environment
variables, which is kinda annoying.
- There might be some performance issues with creating a new
`NushellPrompt` object and cloning the `Stack` for every segment of the
transient prompt. What's more, the transient prompt is added to the
`Reedline` object whether or not the user has enabled transient prompt,
so if there are indeed performance issues, simply disabling the
transient prompt won't help.
- Perhaps instead of a separate `TRANSIENT_PROMPT_INDICATOR_VI_INSERT`
and `TRANSIENT_PROMPT_INDICATOR_VI_NORMAL`, `TRANSIENT_PROMPT_INDICATOR`
could be used for both (if it exists). Insert and normal mode don't
really matter for previously entered commands.
<!--
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.
-->
When parse_range get an item like ((((1..2)))) it would try to parse
"((((1" with a long chain of recursive parsers, namely:
- parse_value
- parse_paren_expr
- parse_full_cell_path
- parse_block
- parse_pipeline
- parse_builtin_commands
- parse_expression
- parse_math_expression
- parse_value
- ...
where `parse_paren_expr` calls `parse_range` in turn. Because at any
time in the chain `parse_paren_expr` can call `parse_range`, which will
then continue the chain, we get quadratic number of function calls, each
linear on the size of the input
By checking with the lexer that the parens are matched, we prevent the
long chain from being called on unmatched braces. Now, this is still
more quadratic than it needs to be, to fix that, we should process
parens only once, instead of on each recursive call
# User-Facing Changes
<!-- List of all changes that impact the user experience here. This
helps us keep track of breaking changes. -->
Speed improvements in some edge cases
# Tests + Formatting
Not sure how to test this, maybe I could add a benchmark
<!--
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.
-->
# Other notes
Found using the fuzzer, by setting a timeout on max run-time. It also
found a stack-overflow on too many parentheses, which this doesn't fix.
# Description
This PR fixes a bug found by @kubouch where `$nu` is not evaluated as a
constant when the `--ide-check` parameter is passed to nushell for the
LSP-like functionality. Not quite sure my changes are proper but it
seems to work!
### Before
### After
# 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.
-->
---------
Co-authored-by: Jakub Žádník <kubouch@gmail.com>
# Description
While reviewing #10350 I noticed that a `HashSet<usize>` was used to
deduplicate the incoming rows, which are then sorted after cloning to a
separate `Vec`. This sounds like a candidate for a `BTreeSet` which
guarantees the ordering.
In the process I removed some dead code.
- Use `BTreeSet` instead of `HashSet`
- Remove dead `skip` logic
- Use `BTreeSet` directly in `NthIterator`
- Consume `BTreeSet` through `Peekable<IntoIter>`
Factor the big parts into separate files:
- `state_delta.rs`
- `state_working_set.rs`
- smaller `usage.rs`
This required adjusting the visibility of several parts.
Makes `StateDelta` transparent for the module.
Trying to reduce visibility in some other places
# Description
This PR should close#10085
Maps `DirectoryNotFound` errors to `FileNotFound`. All other errors are
left unchanged.
# User-Facing Changes
This means a user will see `FileNotFound` instead of `DirectoryNotFound`
which is more meaning full to the user.
# Description
Mention the expected substring in:
- `fn run_test_contains()`
- `fn test_ide_contains()`
- `fn fail_test()`
# Developer-Facing Changes
You can read the test failure to figure out what is going on, don't
necessarily have to look up the test
New look of failure
```
failures:
---- tests::test_math::broken_math stdout ----
stdout:
stderr: Error: nu::parser::incomplete_math_expression
× Incomplete math expression.
╭─[/tmp/.tmpz0V8nC:1:1]
1 │ 3 +
· ┬
· ╰── incomplete math expression
╰────
Expected error to contain: Incomplete universe
thread 'tests::test_math::broken_math' panicked at 'assertion failed: !stderr.is_empty() && stderr.contains(expected)', src/tests.rs:181:5
```
# Description
Unify the logic between `nu!` and `nu_with_std!`.
The inner code actually does not contain any variadic components. So it
can safely be abstracted into a function.
Similarly simplify the variadic to an array in `nu_with_plugin!`
This also seems to simplify the codegen for tests.
Comparing the size of the `/target/debug` folder after running:
```sh
cargo clean --profile dev
cargo build --workspace --tests
```
With this branch a reduction from `8.9GB` to `8.7GB`
# User-Facing Changes
None
# Tests + Formatting
No changes necessary
The *toolchain.toml* doc incorrectly stated how many releases behind
stable we will be...
It said we would be 1 release behind when actually we want to be 2
releases behind stable.
