Currently hardcoded fonts and sizes.

The following tests fail due to missing cell protection:

• ed - Protection missing
• decsed
• decsel
• decsera

The xterm.js-based reflow seems bad, it attempts to reflow what is on the screen, rather than keeping track of the semantics, which means that resizing will throw away data.

This should rather keep track at a character level, whether the position was triggered due to a hard newline or a wrapping.

The issue that is causing trouble are the 8-bit sequnces that can be sent, which conflict with UTF8, in particular the DCS initiator (0x90). Previously the code was consuming this sequence when on 8 bit mode:

            if currentState == .Ground && code > 0x1f
           // Gather print data

So this was consuming the 0x90 as print data. So the variation became:

if currentState == .Ground && (code > 0x1f && code < 0x80 || (code > 0xc2 && code < 0xf3)) {

But this ends up eating UTF8 sequences (see mc running in this mode)

• Add support for passing fonts and color schemes
• Add support for changing font on the fly
• Add support for changing font size on the fly
• Add support for changing color scheme on the fly
• Add support for picking one of the new built-in color ANSI palettes
• Add support for setting the history size.

The above probably should be surfaced as properties. Currently we have options which is ok, but I find cumbersome to copy, update and change. I rather have a more idiomatic set of properties.

Support a way of registering a handler to determine whether there is a url worth clicking on the display.

Print the dirty rect to see.

Selection area doesn't cover selected text in full

Implement Command-R to switch between sending mouse events to target application and capturing it for selection and vice versa, like the Mac terminal.app does

After resizing the window, what happens it seems like I am painting parts from the buffer delta:

This happens in the current mask building code which is wrong:

Inserting text, when it pushes a long string, corrupts the next line.

Repeatedly pressing control-l re-renders things differently.

Correct:

Repeatedly pressing control-l sometimes shows this:

Double click should select the word
Triple click should select the line
On a large buffer, starting the selection, starts from somewhere way back
Scrolling looks busted.

Typing:

"while true foo" esc-b esc-b esc-d

This should delete "true", and it does, but also does not display "foo" anymore.

This happens on first use, afterwards this seems to work.

This seems to also cause, paste this:

while true; date; sleep 0.1; done

Then go with alt-b up until date and type "do", the rest of the line is erased.

https://vt100.net/docs/vt510-rm/DECLRMM.html

VI is not working when you insert text.

Also, one time I saw it crash in a call to splice.

Perhaps all uses of col, row pairs, should be replaced with Position structure.

The reason for this problem is that we turn on the mouse-event sending as a global variable, but there is no way of turning it off - only turning off the individual modes.

So we need to remove the global from the Terminal, and make the MacTerminalView rely on the independent mouse variables in Terminal to operate, and not this ugly global

I suspect there is not enough room between the cells and they are being overdrawn.

Not sure why, this causes the selection to be extended sometimes.

Repro:

in "bash$ foo" double click on the space between "$" and "foo", that should select that space, and quickly move a couple of lines down and quickly click - this currently extends the selection, because it receives a mouseDrag event, even when the mouse is up.

Once the buffer starts to get reused, a simple"

while true; date; sleep 0.1; done will show it.

• apc - s8c1t dependency
• bs - works, but requires terminal to be in 80x25 to test
• cup
• dcs
• cht
• cat
• cha
• chi
• change_color, change_dynamic_color, change_special_color
• cnl
• cpl
• cr
• cub
• cud
• cuf
• cup
• cuu
• da - caveat, we report our own version of the terminal
• DCH
• dcs
• decaln - test pattern tests
• decbi - DECBI_Basic, DECBI_LeftOfMargin, DECBI_Scrolls, DECBI_WholeScreenScrolls
• deccra - modulo a couple of debatable changes
• decdc
• decdsr
• decera
• decfi DECFI_WholeScreenScrolls
• decfra
• decic - DECIC_CursorWithinTopBotton, DECIC_IsNoOpWhenCursorBeginsOutsideScrollRegion
• decid - obsolete
• decrqm - not supported
• decrqss - running in iterm2 mode, and disagree on hardcoded values
• decscl - not worth it?
• decset - mostly there, a few issues remain, and a few ugly things
• decset-tit-inhibit
• decstbm
• decstr - we pass, but we have a "should not work that works"
• dl - we need support for margins in the various delete line operations
• ech - needs protected cells
• el - needs protected cells
• ff
• hpa - delta are bugs fixed in SwiftTerm
• hpr - delta are bugs fixed in SwiftTerm
• hts
• hvp
• ich
• il - scroll region support
• ind - scroll region support
• lf - scroll region support
• nel - scroll region support
• pm
• rep
• resetColor
• resetSpecialColor
• ri - scroll region
• RIS - missing one column switch
• s8c1t - missing cell protection support and see the bug on utf8 sequences and escape parser
• rm
• sd
• sm
• sos
• su
• tbc
• vpa - problem with the origin cursor location
• vpr
• vt - one missing
• xterm_save - low priority?
• xterm_winops lots of goodies here

And this causes problems for the test suite when using the xterm mode.

• does not copy the last character on Copy operation - only on the last visible character of a row.
• Should copy immediately to a buffer, to not rely on the "Terminal" storage, which can go away after scrolling

When the baseline "Terminal" is embedded, the capabilities that it should surface to the client are bound by what the view can do. A cocoa view is very capable, but an ncurses view is not - so I need a way of customizing how much the Terminal reports back when capabilities are requested.

Currently it only expands by letter, and should expand by word

man nc

Then page to the end, then repeatedly hit "b", it crashes here:

        precondition(start < self.count)
with start = 188
count = 111

head -60 /etc/services crashes because we have not filled out the buffer with an NSAttributedString, I think that I should populate on any growth or hook up makeEmpty.

Tests:

• apc -

It tries to ensure a valid command (if things are invalid like left is higher that right), but in these cases, the commands should stop - see test test_DECCRA_invalidSourceRectDoesNothing for example

There is now a work in progress patch to add support for TrueColor in the reworkAttributes brand, this is a list of pending issues:

• In Terminal.swift, the SGR handler needs to implement the support for picking up the RGB data for fg and bg, as well as the indexed colors
• While I am doing indexed colors, complete the support for configuring indexed colors -
• Validate the constants documents in xterm.js vs the spec, because they are conflicting (the values for background have a range of colors, which on the original spec just listed one - 100)
• Update Attribute.toSgr() to render the additional trueColor values.

Currently paints the whole buffer, could just paint the affected rows.

This can protect some characters on the screen

https://vt100.net/docs/vt510-rm/DECSCA.html

• It does not show up solid, with the color I desire - this might be an issue with the default initializer in the property.

• It does not show up after focus out/focus-in

The vttest for some character sets is not setting the second part of the attribute correctly for the second half

Should be: