## Summary of the Pull Request
Move `ICoreSettings` and `IControlSettings` from the TerminalSettings project to the TerminalCore and TerminalControl projects respectively. Also entirely removes the TerminalSettings project.
The purpose of these interfaces is unchanged. `ICoreSettings` is used to instantiate a terminal. `IControlSettings` (which requires an `ICoreSettings`) is used to instantiate a UWP terminal control.
## References
Closes#7140
Related Epic: #885
Related Spec: #6904
## PR Checklist
* [X] Closes#7140
* [X] CLA signed
* [X] Tests ~added~/passed (no additional tests necessary)
* [X] ~Documentation updated~
* [X] ~Schema updated~
## Detailed Description of the Pull Request / Additional comments
A lot of the work here was having to deal with winmd files across all of these projects. The TerminalCore project now outputs a Microsoft.Terminal.TerminalControl.winmd. Some magic happens in TerminalControl.vcxproj to get this to work properly.
## Validation Steps Performed
Deployed Windows Terminal and opened a few new tabs.
This fixes the build the rest of the way in VS 16.7. Something about the
way we were using the Store/Container flags caused some of our projects
to end up linking kernel32.lib only (MinimalCoreWin==KernelOnly).
The best way to solve it once and for all is to make sure MinimalCoreWin
is always set.
References 313568d0e5.
## Summary of the Pull Request
Adds support for two actions, `closeOtherTabs` and `closeTabsAfter`. Both these actions accept an `index` parameter.
* `closeOtherTabs`: Close tabs other than `index`
* `closeTabsAfter`: Close tabs after `index` (This is also "Close tabs to the right")
## References
* This PR is being made to unblock @RahulRavishankar in #1912
## PR Checklist
* [x] I work here
* [ ] Tests added/passed
* [x] Requires documentation to be updated
* [ ] We should file an issue for "add an `index` param to `closeTab`" to add similar support to the close tab action
* [ ] We should file an issue for "make the `index` param to `closeOtherTabs`, `closeTabsAfter` optional" to make them both work on the _active_ tab when there's no `index` provided
## Validation Steps Performed
* _Verified that_ closing all tabs when I have the `index`'th tab selected _works as expected_
* _Verified that_ closing all tabs when I have a tab other than the `index`'th tab selected _works as expected_
* _Verified that_ closing tabs to the right when I have the `index`'th tab selected _works as expected_
* _Verified that_ closing tabs to the right when I have a tab other than the `index`'th tab selected _works as expected_
- This one has one caveat: for whatever reason, if you run this action when the tab that's currently focused is _before_ the `index` param, then the tabs will expand to fill the entire width of the tab row, until you mouse over them. Probably has something to do with tabs not resizing down until there's a mouse exit event.
Up until #4999 we deferred all key events to the character event handler
for which `ToUnicodeEx` returned a valid character and alternatively
those who aren't a special key combination as listed in
`TerminalInput`'s implementation.
Since #4999 we started acknowledging/handling all key events no matter
whether they're actually a known key combination. Given non-ASCII inputs
the Win32 `SendInput()` method generates certain sequences that aren't
recognizable combinations though and if they're handled by the key event
handler no follow up character event is sent containing the unicode
character.
This PR adds another condition and defers all key events without scan
code (i.e. those not representable by the current keyboard layout) to
the character event handler.
I'm absolutely not certain that this PR doesn't have a negative effect
on other kinds of inputs.
Is it common for key events to not contain a scan code? I personally
haven't seen it happen before AutoHotKey/SendInput.
Before this PR is merged it'd be nice to have a good testing plan in
place in order to ensure nothing breaks.
## Validation Steps Performed
Remapped `AltGr+8` to `»` using AutoHotKey using `<^>!8::SendInput {Raw}»`.
Ensured `»` is printed if `AltGr+8` is pressed.
Closes#7064Closes#7120
Carlos Zamora (1)
* UIA: use full buffer comparison in rects and endpoint setter (GH-6447)
Dan Thompson (2)
* Tweaks: normalize TextAttribute method names (adjective form) (GH-6951)
* Fix 'bcz exclusive' typo (GH-6938)
Dustin L. Howett (4)
* Fix VT mouse capture issues in Terminal and conhost (GH-7166)
* version: bump to 1.3 on master
* Update Cascadia Code to 2007.15 (GH-6958)
* Move to the TerminalDependencies NuGet feed (GH-6954)
James Holderness (3)
* Render the SGR "underlined" attribute in the style of the font (CC-7148)
* Add support for the "crossed-out" graphic rendition attribute (CC-7143)
* Refactor grid line renderers with support for more line types (CC-7107)
Leonard Hecker (1)
* Added til::spsc, a lock-free, single-producer/-consumer FIFO queue (CC-6751)
Michael Niksa (6)
* Update TAEF to 10.57.200731005-develop (GH-7164)
* Skip DX invalidation if we've already scrolled an entire screen worth of height (GH-6922)
* Commit attr runs less frequently by accumulating length of color run (GH-6919)
* Set memory order on slow atomics (GH-6920)
* Cache the viewport to make invalidation faster (GH-6918)
* Correct comment in this SPSC test as a quick follow up to merge.
Related work items: MSFT-28208358
I found this while crawling through conhost's WindowIo. Mouse wheel
events come in in screen coordinates, unlike literally every other mouse
event.
The WPF control was doing it wrong.
This pull request fixes capture and event generation in VT mouse mode
for both conhost and terminal.
Fixes#6401.
[1/3] Terminal: clamp mouse events to the viewport, don't throw them away
gnome-terminal (at least) sends mouse events whose x/y are at the
extreme ends of the buffer when a drag starts inside the terminal and
then exits it.
We would previously discard any mouse events that exited the borders of
the viewport. Now we will keep emitting events where X/Y=0/w/h.
[2/3] conhost: clamp VT mouse to viewport, capture pointer
This is the same as (1), but for conhost. conhost wasn't already
capturing the pointer when VT mouse mode was in use. By capturing, we
ensure that events that happen outside the screen still result in events
sent to an application (like a release after a drag)
[3/3] wpf: capture the pointer when VT mouse is enabled
This is the same as (2), but for the WPF control. Clamping is handled
in TerminalCore in (1), so we didn't need to do it in WPF.
Move TerminalSettings object from TerminalSettings project
(Microsoft.Terminal.Settings) to TerminalApp project. `TerminalSettings`
specifically operates as a bridge that exposes any necessary information
to a TerminalControl.
Closes#7139
Related Epic: #885
Related Spec: #6904
## PR Checklist
* [X] Closes#7139
* [X] CLA signed
* [X] Tests ~added~/passed (no additional tests necessary)
* [X] ~Documentation updated~
* [X] ~Schema updated~
## Validation Steps Performed
Deployed Windows Terminal and opened a few new tabs.
Add some user research to determine what the average number of characters a user types before executing a cmdpal action.
This might need to be modified when it merges with #6732
Updates TAEF to 10.57.200731005-develop
## PR Checklist
* [x] Helps #6992 by bringing `wttlog.dll` along with the rest of TAEF.
* [x] I work here.
* [x] Automated tests in CI
* [x] No doc/schema update necessary (checked for docs in this repo)
* [x] Am core contributor.
This PR updates the rendering of the _underlined_ graphic rendition
attribute, using the style specified in the active font, instead of just
reusing the grid line at the bottom of the character cell.
* Support for drawing the correct underline effect in the grid line
renderer was added in #7107.
There was already an `ExtendedAttributes` flag defined for the
underlined state, but I needed to update the `SetUnderlined` and
`IsUnderlined` methods in the `TextAttribute` class to use that flag
now in place of the legacy `LVB_UNDERSCORE` attribute. This enables
underlines set via a VT sequence to be tracked separately from
`LVB_UNDERSCORE` grid lines set via the console API.
I then needed to update the `Renderer::s_GetGridlines` method to
activate the `GridLines::Underline` style when the `Underlined`
attribute was set. The `GridLines::Bottom` style is still triggered by
the `LVB_UNDERSCORE` attribute to produce the bottom grid line effect.
Validation
----------
Because this is a change from the existing behaviour, certain unit tests
that were expecting the `LVB_UNDERSCORE` to be toggled by `SGR 4` and
`SGR 24` have now had to be updated to check the `Underlined` flag
instead.
There were also some UI Automation tests that were checking for `SGR 4`
mapping to `LVB_UNDERSCORE` attribute, which I've now substituted with a
test of the `SGR 53` overline attribute mapping to
`LVB_GRID_HORIZONTAL`. These tests only work with legacy attributes, so
they can't access the extended underline state, and I thought a
replacement test that covered similar ground would be better than
dropping the tests altogether.
As far as the visual rendering is concerned, I've manually confirmed
that the VT underline sequences now draw the underline in the correct
position and style, while grid lines output via the console API are
still displayed in their original form.
Closes#2915
This PR adds support for the ANSI _crossed-out_ graphic rendition
attribute, which is enabled by the `SGR 9` escape sequence.
* Support for the escape sequences and storage of the attribute was
originally added in #2917.
* Support for drawing the strikethrough effect in the grid line renderer
was added in #7107.
Since the majority of the code required for this attribute had already
been implemented, it was just a matter of activating the
`GridLines::Strikethrough` style in the `Renderer::s_GetGridlines`
method when the `CrossedOut` attribute was set.
VALIDATION
There were already some unit tests in place in `VtRendererTest` and the
`ScreenBufferTests`, but I've also now extended the SGR tests in
`AdapterTest` to cover this attribute.
I've manually confirmed the first test case from #6205 now works as
expected in both conhost and Terminal.
Closes#6205
When we added support for win32 input mode, we neglected to pass
`ENHANCED_KEY` through the two surfaces that would generate events. This
broke arrow keys in much the same way was #2397, but in a different
layer.
While I was working on the WPF control, I took a moment to refactor the
message cracking out into a helper. It's a lot easier on the eyes than
four lines of bit shifting repeated three times.
Fixes#7074
Allows splitting pane (with default settings) by holding down ALT and pressing the new tab button ('+')
## PR Checklist
* [X] Closes#6757
* [X] Works here.
* [X] Manual test (below)
* [X] Is core contributor.
## More detailed description
Pretty much exactly the code added in #5928 (all credit to @carlos-zamora), but put at the new tab button event binding
## Validation steps
Seems to work - holding ALT while pressing '+' opens a pane instead of a tab. Holding ALT while starting up terminal for the first time does not seem to affect the behaviour.
Retrieved from https://microsoft.visualstudio.com os.2020 OS official/rs_onecore_dep_uxp 29b1a1d663d0047dc0d2125dd05f75959bca27ef
Related work items: MSFT:27866336
This is a refactoring of the grid line renderers, adjusting the line
widths to scale with the font size, and optimising the implementation to
cut down on the number of draw calls. It also extends the supported grid
line types to include true underlines and strike-through lines in the
style of the active font.
The main gist of the optimisation was to render the horizontal lines
with a single draw call, instead of a loop with lots of little strokes
joined together. In the case of the vertical lines, which still needed
to be handled in a loop, I've tried to move the majority of static
calculations outside the loop, so there is bit of optimisation there
too.
At the same time this code was updated to support a variable stroke
width for the lines, instead of having them hardcoded to 1 pixel. The
width is now calculated as a fraction of the font size (0.025 "em"),
which is still going to be 1 pixel wide in most typical usage, but will
scale up appropriately if you zoom in far enough.
And in preparation for supporting the SGR strike-through attribute, and
true underlines, I've extended the grid line renders with options for
handling those line types as well. The offset and thickness of the lines
is obtained from the font metrics (rounded to a pixel width, with a
minimum of one pixel), so they match the style of the font.
VALIDATION
For now we're still only rendering grid lines, and only the top and
bottom lines in the case of the DirectX renderer in Windows Terminal. So
to test, I hacked in some code to force the renderer to use all the
different options, confirming that they were working in both the GDI and
DirectX renderers.
I've tested the output with a number of different fonts, comparing it
with the same text rendered in WordPad. For the most part they match
exactly, but there can be slight differences when we adjust the font
size for grid alignment. And in the case of the GDI renderer, where
we're working with pixel heights rather than points, it's difficult to
match the sizes exactly.
This is a first step towards supporting the strike-through attribute
(#6205) and true underlines (#2915).
Closes#6911
We've been trying to improve the copy/paste experience with the terminal
in Visual Studio. Once of our problematic scenarios is when the terminal
is connected to a remote environment and the user attempts to
copy/paste. This gets forwarded to the remote shell that copy/paste into
the remote clipboard instead of the local one.
So we opted to add ctrl+shift+c/v to the terminal and need access to the
selected text via the terminal control
VALIDATION
Tested with Visual Studio integrated terminal
- Move to GSL 3.1.0 (GH-6908)
- Replace the color table init code with two const arrays (GH-6913)
- Replace basic_string_view<T> with span<const T> (GH-6921)
- Replace gsl::at with a new til::at(span) for pre-checked bounds (GH-6925)
Related work items: MSFT:27866336
[Git2Git] Git Train: Merge of building/rs_onecore_dep_uxp/200720-1445 into official/rs_onecore_dep_uxp
Retrieved from https://microsoft.visualstudio.com os.2020 OS official/rs_onecore_dep_uxp 5e1509a00f1acdefe1e6392468ffc9ae44dc89cb
Related work items: MSFT:25731998
* send alt/F10 through the control
We were not listening for WM_SYSKEY{UP,DOWN}
* extract the actual scancode during WM_CHAR, not the bitfield
We were accidentally sending some of the additional keypress data in with
the character event in Win32 Input Mode
* set default fg/bg to campbell
The WPF control starts up in PowerShell blue even though it's not typically used
in PowerShell blue.
* don't rely on the font to determine wideness
This is a cross-port of #2928 to the WPF control
* deterministic shutdown
In testing, I saw a handful of crashes on teardown because we were not shutting
down the render thread properly.
* don't pass 10 for the font weight ...
When Cascadia Code is set, it just looks silly.
* trigger render when selection is cleared, do it under lock
Fixes#6966.
In UiaTextRange, `_getBufferSize` returns an optimized version of the
size of the buffer to be the origin and the last character in the
buffer. This is to improve performance on search or checking if you are
currently on the last word/line.
When setting the endpoint and drawing the bounding rectangles, we should
be retrieving the true buffer size. This is because it is still possible
to create UiaTextRanges that are outside of this optimized size. The
main source of this is `ExpandToEnclosingUnit()` when the unit is
`Document`. The end _should_ be the last visible character, but it isn't
because that would break our tests.
This is an incomplete solution. #6986 is a follow up to completely test
and implement the solution.
The crash in #6402 was caused by getting the document range (a range of
the full text buffer), then moving the end by one character. When we
get the document range, we get the optimized size of the buffer (the
position of the last character). Moving by one character is valid
because the buffer still has more to explore. We then crash from
checking if the new position is valid on the **optimized size**, not the
**real size**.
REFERENCES
#6986 - follow up to properly handle/test this "end of buffer" problem
Closes#6402
Original notes from @M-Pixel:
> Console applications assume that backgrounds are black, and that
> `lightBlack`/`DarkGrey` are lighter than `black`/`Black`. This
> assumption is accounted for by all color schemes in `defaults.json`,
> except for the Solarized themes.
>
> The Solarized Dark theme, in particular, makes `-Parameters` invisible
> against the background in PowerShell, which is obviously an unacceptable
> usability flaw.
>
> This change makes `black` and `background` to the same (which is common
> throughout the color schemes), and makes `brightBlack` (`DarkGray` in
> .NET) lighter than black (which is obviously more correct given the
> meanings of those words).
Out of the box, we ship a pretty bad behavior.
If I look at all of the existing shipped color schemes--and that
includes things like Tango and One Half--we are universally following a
`background` == `black` rule.
If I consult gnome-terminal or xterm, they do the same thing; Xterm by
default, gnome-terminal for solarized. The background generally matches
color index `0` across all their **dark** schemes. Konsole and
lxterminal disagree and map background to `0 intense` for Solarized.
I want to put our Solarized schemes on a deprecation path, but
unfortunately we still need to ship _something_ for users who we're
going to strand on them.
I'm going to have to swallow my bitter and say that yes, we should
probably just change the index mapping and go with something that works
right out of the box while we figure out how to do perceptual color
nudging and eventually remove bad defaults (like Solarized).
From #6618.
Fixes#4047.
Closes#6618.
## Summary of the Pull Request
Adds a execute commandline action (`wt`), which lets a user bind a key to a specific `wt` commandline. This commandline will get parsed and run _in the current window_.
## References
* Related to #4472
* Related to #5400 - I need this for the commandline mode of the Command Palette
* Related to #5970
## PR Checklist
* [x] Closes oh, there's not actually an issue for this.
* [x] I work here
* [x] Tests added/passed
* [ ] Requires documentation to be updated - yes it does
## Detailed Description of the Pull Request / Additional comments
One important part of this change concerns how panes are initialized at runtime. We've had some persistent trouble with initializing multiple panes, because they rely on knowing how big they'll actually be, to be able to determine if they can split again.
We previously worked around this by ignoring the size check when we were in "startup", processing an initial commandline. This PR however requires us to be able to know the initial size of a pane at runtime, but before the parents have necessarily been added to the tree, or had their renderer's set up.
This led to the development of `Pane::PreCalculateCanSplit`, which is very highly similar to `Pane::PreCalculateAutoSplit`. This method attempts to figure out how big a pane _will_ take, before the parent has necessarily laid out.
This also involves a small change to `TermControl`, because if its renderer hasn't been set up yet, it'll always think the font is `{0, fontHeight}`, which will let the Terminal keep splitting in the x direction. This change also makes the TermControl set up a renderer to get the real font size when it hasn't yet been initialized.
## Validation Steps Performed
This was what the json blob I was using for testing evolved into
```json
{
"command": {
"action":"wt",
"commandline": "new-tab cmd.exe /k #work 15 ; split-pane cmd.exe /k #work 15 ; split-pane cmd.exe /k media-commandline ; new-tab powershell dev\\symbols.ps1 ; new-tab -p \"Ubuntu\" ; new-tab -p \"haunter.gif\" ; focus-tab -t 0",
},
"keys": ["ctrl+shift+n"]
}
```
I also added some tests.
# TODO
* [x] Creating a `{ "command": "wt" }` action without a commandline will spawn a new `wt.exe` process?
- Probably should just do nothing for the empty string
We spend a lot of time invalidating in the DX Renderer. This is a
creative trick to not bother invalidating any further if we can tell
that the bitmap is already completely invalidated. That is, if we've
scrolled at least an entire screen in height... then the entire bitmap
had to have been marked as invalid as the new areas were "uncovered" by
the `InvalidateScroll` command. So further setting invalid bits on top
of a fully invalid map is pointless.
Note: I didn't use `bitmap::all()` here because it is significantly
slower to check all the bits than it is to just reason out that the
bitmap was already fully marked.
## Validation Steps Performed
- Run `time cat big.txt`. Checked average time before/after, WPR traces
before/after.
The act of calling `InsertAttrRuns` is relatively slow. Instead of
calling it a bunch of times to meddle with colors one cell at a time,
we'll accumulate a length of color and call it to make it act all at
once. This is great for when one color full line is getting replaced
with another color full line OR when a line is being replaced with the
same color all at once. There's significantly fewer checks to be made
inside `InsertAttrRuns` if we can help it out by accumulating the length
of each color before asking it to stitch it into the storage.
Validation
----------
- Run `time cat big.txt` and `time cat ls.txt` under VS Performance
Profiler.
By default, the memory order on atomics is `seq_cst`. This is a relatively expensive ordering and it shows in situations where we're rapidly signaling a consumer to pick up something from a producer. I've instead attempted to switch these to `release` (producer) and `acquire` (consumer) to improve the performance of these signals.
## Validation Steps Performed
- Run `time cat big.txt` and `time cat ls.txt` under VS Performance Profiler.
## PR Checklist
* [x] Closes perf itch
* [x] I work here
* [x] Manual test
* [x] Documentation irrelevant.
* [x] Schema irrelevant.
* [x] Am core contributor.
## Summary of the Pull Request
Text can have various attributes, such as "bold", "italic", "underlined", etc. The TextAttribute class embodies this. It has methods to set/query these attributes.
This change tweaks a few of the method names to make them match. I.e. for an imaginary text property "Foo", we should have methods along the lines of:
```
IsFoo
SetFoo(bool isFoo)
```
And variations should match: we should have "Foo" and "OverFoo", not "Fooey" and "OverFoo".
I chose to standardize on the adjective form, since that's what we are closest to already. The attributes I attacked here are:
SetItalic**s** --> SetItalic
SetUnderline --> SetUnderline**d**
SetOverline --> SetOverline**d**
("italic" is an adjective; "italics" is a plural noun, representing letters or words in an italic typeface)
And I also added methods for "DoublyUnderlined" for good measure.
I stopped short of renaming the GraphicsOptions enum values to match, too; but I'd be willing to do that in a follow-up change if people wanted it.
## Validation Steps Performed
It builds, and tests still pass.
This pull request converts the following JSON deserializers to use the
new JSON deserializer pattern:
* Profile
* Command
* ColorScheme
* Action/Args
* GlobalSettings
* CascadiaSettingsSerialization
This is the completion of a long-term JSON refactoring that makes our
parser and deserializer more type-safe and robust. We're finally able to
get rid of all our manual enum conversion code and unify JSON conversion
around _types_ instead of around _keys_.
I've introduced another file filled with template specializations,
TerminalSettingsSerializationHelpers.h, which comprises a single unit
that holds all of the JSON deserializers (and eventually serializers)
for every type that comes from TerminalApp or TerminalSettings.
I've also moved some types out of Profile and GlobalAppSettings into a
new SettingsTypes.h to improve settings locality.
This does to some extent constitute a breaking change for already-broken
settings. Instead of parsing "successfully" (where invalid values are
null or 0 or unknown or unset), deserialization will now fail when
there's a type mismatch. Because of that, some tests had to be removed.
While I was on a refactoring spree, I removed a number of helpless
helpers, like GetWstringFromJson (which converted a u8 string to an
hstring to make a wstring out of its data pointer :|) and
_ConvertJsonToBool.
In the future, we can make the error types more robust and give them
position and type information such that a conformant application can
display rich error information ("line 3 column 3, I expected a string,
you gave me an integer").
Closes#2550.
The command palette is a ListView of commands. As you type into the
search box, commands are added or removed from the ListView. Currently,
each update is done by completely clearing the backing list, then adding
back any items that should be displayed. However, this defeats the
ListView's built-in animations: upon every keystroke, ListView displays
its list-clearing animation, then animates the insertion of every item
that wasn't deleted. This results in noticeable flickering.
This PR changes the update logic so that it updates the list using
(roughly) the minimum number of Insert and Remove calls, so the ListView
makes smoother transitions as you type.
I implemented it by keeping the existing code that builds the filtered
list, but I changed it to build into a scratch list. Then I grafted on
a generic delta algorithm to make the real list look like the scratch
list.
To verify the delta algorithm, I tested all 360,000 permutations of
pairs of up to 5 element lists in a toy C# app.
## Validation
I'm not sure if my screen capture tool really caught all the flickering
here, but the screencasts below should give a rough idea of the
difference. (All the flickering was becoming a nuisance while I was
testing out the HC changes.)
See the images in #6939 for more info.
Co-authored-by: Jeffrey Tippet <jtippet@microsoft.com>
In `Renderer::TriggerRedraw`, the act of fetching the viewport from the
`pData` over and over is wasted time. We already have a cached variable
of the viewport that is updated on every scroll check (on
`TriggerScroll` and on `PaintFrame`.) Scrolling wouldn't be working
correctly if the clients weren't already notifying us that the viewport
has changed for scroll purposes, so we can just keep using that cached
value for the invalidation restriction to speed things up over fetching
it again.
## Validation Steps Performed
- Run `time cat big.txt`. Checked average time before/after, WPR traces
before/after.
## PR Checklist
* [x] Closes perf itch
* [x] I work here
* [x] Manual test
* [x] Documentation irrelevant.
* [x] Schema irrelevant.
* [x] Am core contributor.
## Summary of the Pull Request
This PR adds the `til::spsc` namespace, which implements a lock-free, single-producer, single-consumer FIFO queue ("channel"). The queue efficiently blocks the caller using Futexes if no data can be written to / read from the queue (e.g. using `WaitOnAddress` on Windows). Furthermore it allows batching of data and contains logic to signal the caller if the other side has been dropped/destructed.
## PR Checklist
* [ ] Closes #xxx
* [x] CLA signed. If not, go over [here](https://cla.opensource.microsoft.com/microsoft/Terminal) and sign the CLA
* [x] Tests added/passed
* [ ] Documentation updated. If checked, please file a pull request on [our docs repo](https://github.com/MicrosoftDocs/terminal) and link it here: #xxx
* [ ] Schema updated.
* [x] I've discussed this with core contributors already. If not checked, I'm ready to accept this work might be rejected in favor of a different grand plan. Issue number where discussion took place: #xxx
## Detailed Description of the Pull Request / Additional comments
`til::spsc::details::arc<T>` contains most of the queue's logic and as such has the relevant documentation for its design.
## Validation Steps Performed
The queue was tested on Windows, Linux and macOS using MSVC, gcc and llvm and each of their available runtime introspection utilities in order to ensure no race conditions or memory leaks occur.
The recent changes to use gsl::span everywhere added a few bounds checks
along codepaths where we were already checking bounds. Some of them may
be non-obvious to the optimizer, so we can now use til::at to help them
along.
To accomplish this, I've added a new overload of til::at that takes a
span and directly accesses its backing buffer.
We were using std::basic_string_view as a stand-in for std::span so that
we could change over all at once when C++20 dropped with full span
support. That day's not here yet, but as of 54a7fce3e we're using GSL 3,
whose span is C++20-compliant.
This commit replaces every instance of basic_string_view that was not
referring to an actual string with a span of the appropriate type.
I moved the `const` qualifier into span's `T` because while
`basic_string_view.at()` returns `const T&`, `span.at()` returns `T&`
(without the const). I wanted to maintain the invariant that members of
the span were immutable.
* Mechanical Changes
* `sv.at(x)` -> `gsl::at(sp, x)`
* `sv.c{begin,end}` -> `sp.{begin,end}` (span's iterators are const)
I had to replace a `std::basic_string<>` with a `std::vector<>` in
ConImeInfo, and I chose to replace a manual array walk in
ScreenInfoUiaProviderBase with a ranged-for. Please review those
specifically.
This will almost certainly cause a code size regression in Windows
because I'm blowing out all the PGO counts. Whoops.
Related: #3956, #975.
This results in smaller code and faster copying. I chose til::color even
though it results in slightly worse codegen (byteswapping in a tight
loop) than COLORREF (SSE-enlightened block copy) because eventually the
internal representations of the color tables will also be til::color and
_then_ it will become a block copy.
This PR adds support for always on top mode, via two mechanisms:
* The global setting `alwaysOnTop`. When set to true, the window will be
created in the "topmost" group of windows. Changing this value will
hot-reload whether the window is in the topmost group.
* The action `toggleAlwaysOnTop`, which will toggle the `alwaysOnTop`
property at runtime.
## Detailed Description of the Pull Request / Additional comments
All "topmost" windows maintain an internal z-ordering relative to one
another, but they're all always above all other "non-topmost" windows.
So multiple Windows Terminal windows which are both `alwaysOnTop` will
maintain a z-order relative to one another, but they'll all be on top of
all other windows.
## Validation Steps Performed
Toggled always on top mode, both in the settings and also at runtime,
and verified that it largely did what I expected.
Closes#3038
This adds `nt`, `sp`, and `ft` as aliases for `new-tab`, `split-pane`,
and `focus-tab`, respectively. These do exactly the same thing as their
long for counterparts, but are just shorter, for those of us who type
slower than a fifth grader 👀
Now you can do
```
wt nt cmd.exe /k #work 15 ; sp cmd.exe /k #work 15 ; sp cmd.exe /k
media-commandline ; nt powershell dev\\symbols.ps1 ; nt -p \"Ubuntu\" ;
nt -p \"Ubuntu\" ; ft -t 0
```
instead of
```
new-tab cmd.exe /k #work 15 ; split-pane cmd.exe /k #work 15 ;
split-pane cmd.exe /k media-commandline ; new-tab powershell
dev\\symbols.ps1 ; new-tab -p \"Ubuntu\" ; new-tab -p \"Ubuntu\" ;
focus-tab -t 0
```
The pattern I'm using here is that each of these subcommands now has a
little helper lambda that actually sets up the subcommand with the
required arguments, and we just call that lambda twice, once for the
long-form of the command, and again for the short.
I imagine that in the future, we won't necessarily have short-forms for
every subcommands, so if there are future conflicts we'd have to figure
that out pre-emptively, but these all seem like they'll need a short
form.
Closes#5466
GSL 3, the next major version of GSL after the one we're using, replaced
their local implementation of `span` with one that more closely mimics
C++20's span. Unfortunately, that is a breaking change for all of GSL's
consumers.
This commit updates our use of span to comply with the new changes in
GSL 3.
Chief among those breaking changes is:
* `span::at` no longer exists; I replaced many instances of `span::at`
with `gsl::at(x)`
* `span::size_type` has finally given up on `ptrdiff_t` and become
`size_t` like all other containers
While I was here, I also made the following mechanical replacements:
* In some of our "early standardized" code, we used std::optional's
`has_value` and `value` back-to-back. Each `value` incurs an
additional presence test.
* Change: `x.value().member` -> `x->member` (`optional::operator->`
skips the presence test)
* Change: `x.value()` -> `*x` (as above)
* GSL 3 uses `size_t` for `size_type`.
* Change: `gsl::narrow<size_t>(x.size())` -> `x.size()`
* Change: `gsl::narrow<ptrdiff_t>(nonSpan.size())` -> `nonSpan.size()`
during span construction
I also replaced two instances of `x[x.size() - 1]` with `x.back()` and
one instance of a manual array walk (for comparison) with a direct
comparison.
NOTE: Span comparison and `make_span` are not part of the C++20 span
library.
Fixes#6251
## Summary of the Pull Request
This PR adds support for the `SGR 8` and `SGR 28` escape sequences,
which enable and disable the _concealed/invisible_ graphic rendition
attribute. When a character is output with this attribute set, it is
rendered with the same foreground and background colors, so the text is
essentially invisible.
## PR Checklist
* [x] Closes#6876
* [x] CLA signed.
* [x] Tests added/passed
* [ ] Documentation updated.
* [ ] Schema updated.
* [x] I've discussed this with core contributors already. Issue number
where discussion took place: #6876
## Detailed Description of the Pull Request / Additional comments
Most of the framework for this attribute was already implemented, so it
was just a matter of updating the `TextAttribute::CalculateRgbColors`
method to make the foreground the same as the background when the
_Invisible_ flag was set. Note that this has to happen after the
_Reverse Video_ attribute is applied, so if you have white-on-black text
that is reversed and invisible, it should be all white, rather than all
black.
## Validation Steps Performed
There were already existing SGR unit tests covering this attribute in
the `ScreenBufferTests`, and the `VtRendererTest`. But I've added to the
`AdapterTest` which verifies the SGR sequences for setting and resetting
the attribute, and I've extended the `TextAttributeTests` to verify that
the color calculations return the correct values when the attribute is
set.
I've also manually confirmed that we now render the _concealed text_
values correctly in the _ISO 6429_ tests in Vttest. And I've manually
tested the output of _concealed_ when combined with other attributes,
and made sure that we're matching the behaviour of most other terminals.
This parameter was added as a workaround for our fast trackpad
scrolling. Since that was fixed before 1.0 shipped, in #4554, it has
been largely vestigial. There is no reason for us to keep it around any
longer.
It was also the only "logic" in TerminalSettings, which is otherwise a
library that only transits data between two other libraries.
I have not removed it from the schema, as I do not want to mark folks'
settings files invalid to a strict schema parser.
While I was in the area, I added support for "scroll one screen at a
time" (which is represented by the API returning WHEEL_PAGESCROLL),
fixing #5610. We were also storing it in an int (whoops) instead of a
uint.
Fixes#5610
console: switch to /Zc:wchar_t (native wchar_t)
This matches what we use in OpenConsole and makes {fmt} play nice.
I've also removed the workaround we introduced into OutputCellIterator
to work around not using /Zc:wchar_t.
Fixes MSFT:27626309.
Fixes GH-2673.
Retrieved from https://microsoft.visualstudio.com os.2020 OS official/rs_onecore_dep_uxp 1508f7c232ec58bebc37fedfdec3eb8f9bff5502
## Summary of the Pull Request
This PR adds support for the `SGR 2` escape sequence, which enables the
ANSI _faint_ graphic rendition attribute. When a character is output
with this attribute set, it uses a dimmer version of the active
foreground color.
## PR Checklist
* [x] Closes#6703
* [x] CLA signed.
* [x] Tests added/passed
* [ ] Documentation updated.
* [ ] Schema updated.
* [x] I've discussed this with core contributors already. Issue number where discussion took place: #6703
## Detailed Description of the Pull Request / Additional comments
There was already an `ExtendedAttributes::Faint` flag in the
`TextAttribute` class, but I needed to add `SetFaint` and `IsFaint`
methods to access that flag, and update the `SetGraphicsRendition`
methods of the two dispatchers to set the attribute on receipt of the
`SGR 2` sequence. I also had to update the existing `SGR 22` handler to
reset _Faint_ in addition to _Bold_, since they share the same reset
sequence. For that reason, I thought it a good idea to change the name
of the `SGR 22` enum to `NotBoldOrFaint`.
For the purpose of rendering, I've updated the
`TextAttribute::CalculateRgbColors` method to return a dimmer version of
the foreground color when the _Faint_ attribute is set. This is simply
achieved by dividing each color component by two, which produces a
reasonable effect without being too complicated. Note that the _Faint_
effect is applied before _Reverse Video_, so if the output it reversed,
it's the background that will be faint.
The only other complication was the update of the `Xterm256Engine` in
the VT renderer. As mentioned above, _Bold_ and _Faint_ share the same
reset sequence, so to forward that state over conpty we have to go
through a slightly more complicated process than with other attributes.
We first check whether either attribute needs to be turned off to send
the reset sequence, and then check if the individual attributes need to
be turned on again.
## Validation
I've extended the existing SGR unit tests to cover the new attribute in
the `AdapterTest`, the `ScreenBufferTests`, and the `VtRendererTest`,
and added a test to confirm the color calculations when _Faint_ is set
in the `TextAttributeTests`.
I've also done a bunch of manual testing with all the different VT color
types and confirmed that our output is comparable to most other
terminals.
