Currently (0.9.15), It is not cleared, but the indicator is not shown when the mnemonic size is changed.
Which is better, cleared or displayed ...

T/O

This is seen at least from v1.0.6.

The assemble list isn't shown until it is scrolled after "disassemble" button clicked.

It must be able to do with arrows and pageUp/Dn.

In instructions like LD A,(IX+d), the emulator considers the 8-bit displacement d as an unsigned integer, ranging from 0 up to and including 255.

This is incorrect. A real Z80 considers d as a signed integer, ranging from -128 up to and including +127. It is typically sign-extended to 16 bits before being added to IX or IY. This is similar to the way e is added to the program counter in relative jumps like JR e.

For example, DD 7E FF is the object code for LD A,(IX-1). The emulator incorrectly interprets it as LD A,(IX+255).

This is a serious dealbreaker for programs that rely on indexed addressing.

The issue is easy to reproduce with the following program. It should print digits 123456789 at the top left of the screen. In the emulator, digit 9 is printed on the right spot, all others are printed in the center of the screen.

0000   DD 21 59 D0            LD   IX,0D059h   
0004   DD 36 00 29            LD   (IX+0),29h   
0008   DD 36 FF 28            LD   (IX-1),28h   
000C   DD 36 FE 27            LD   (IX-2),27h   
0010   DD 36 FD 26            LD   (IX-3),26h   
0014   DD 36 FC 25            LD   (IX-4),25h   
0018   DD 36 FB 24            LD   (IX-5),24h   
001C   DD 36 FA 23            LD   (IX-6),23h   
0020   DD 36 F9 22            LD   (IX-7),22h   
0024   DD 36 F8 21            LD   (IX-8),21h   
0028   76                     HALT

Different characters are shown in the dump list, when the radio button is changed after scrolled.

When the width of the dump list is to narrow, the header row must be synchronized to the body scrolling by its horizontal scrollbar.

For instance:

• The title music of 'Wonder House' is not correct.
• The 'Building Hopper' never start the game.

But the MZT file downloaded from CMT section is correct.

If the assemble list of stopped position does not exist, it should be disassembled automatically.

In hovering, the droppable target elements are disappeared.
This feature might not be needed.

T/O

Don't work Speed Control. Always Full Speed.

Currently, the object's key is an address and the data is VRAM information.
But the addresses must be enumerated from the object and converted to a number.
I think it is slow.
So notification data would be better to be an array.

I think that It might be related to the worker side rendering.

This occured after the first message screen of running WONDER HOUSE was closed.
The stack pointer seems that takes negative value at the time.

When the container panel width is changed to narrow, the UI-controls to specify an address for the memory dump list are folded to next line and the layout is crumbled.

While testing another issue (#161), I noticed something weird. When trying to load a very small program, the emulated CPU will stop at whatever instruction it was busy with before loading the new program. This could be an instruction in the monitor's keyboard handler, but it could also be an instruction of the currently running program that had been loaded before loading the new program. So it would appear the problem does not occur while executing the new program, but earlier, possibly while loading the new program into memory.

This seems to happen only with programs of 11 instructions or less. Programs with 12 instructions or more are fine. The number of bytes does not appear to matter; it's really about the number of instructions.

Workaround: pad your program with NOP instructions.

Steps to reproduce:

1. Make an MZT file from the assembly program below.
2. Drag and drop the MZT file onto the emulator.
3. Notice that digits appear on screen - OK.
4. Remove the NOP instructions. Make a new MZT file.
5. Restart the emulator. Drag and drop the MZT file onto the emulator.
6. Notice the 'cmt' name is not updated - not OK.
7. Notice the emulator freezes on some instruction in the monitor - not OK.

0000   DD 21 4A D1            LD   IX,0D000H+330   
0004   DD 36 7E 29            LD   (IX+126),29h   
0008   DD 36 7F 28            LD   (IX+127),28h   
000C   DD 36 80 27            LD   (IX-128),27h   
0010   DD 36 81 26            LD   (IX-127),26h   
0014   DD 36 82 25            LD   (IX-126),25h   
0018   76                     HALT   
0019   00                     NOP   
001A   00                     NOP   
001B   00                     NOP   
001C   00                     NOP   
001D   00                     NOP   

After the mouse enters over the screen once, it closes by timeout.

This is an emulator bug. The MZT is running correctly on an actual machine.

Though the buttons of stop/start and step are located near of the reset button, but those should be near of the assemble list at the right pane. The list has a feature to set break-points and they are used to debug the emulator.

It's inconvenient that the buttons were disappeared after certain time is past when debugging.
And essentially, those buttons are not peripherals of real MZ-700.

It is waste of the screen space.
I want to hide it after few seconds.

I want to be able to use files on cloud owned by individual user.
The format of files are MZT, Z80 assembly source, binary and so on.

The first target cloud that currently I'm planning is Google Documents.
And Dropbox, Microsoft OneNote, GitHub, ... might be followed.

The emulation of the intel8253 is not correct.
CH2 counter of the 8253 never be raising an interrupt.
But, the instruction EI of Z80 is calling address 0038h as interrupt routine.

Feature.

I'm designing a hardware addon fpr MZ-700 here . The purpose is to add 128/256/512KByte through a nvRAM or a NOR-FLASH as a set of 4KByte pages accessible in address range $0000-$0FFF.

To do so, I implement a page switching which is executing an IN instruction in the I/O port $FF. While $FE and $FF are used for printer port, the process of reading at $FF is not implemented for the printer (no device will output data bus). So I use that fact so my device will implement a page selection when reading $FF by using a trick where the content of A through IN AL,($FF) or B through IN A,(C) will be put in address bus A15..A8 and be captured by my device.

To select a page and copy a block of it into RAM:

    LD HL,<first_byte_offset_in_page> 
    LD DE,<address_in_ram>
    LD BC,<last_byte_offset_in_page>-<first_byte_offset_in_page>+1 
    LD A,<page_number>
    DI ; disable interrupts
    IN A,($FF) ; select the page 
    LDIR ; copy a block from the page into RAM
    XOR A ; page 0 (MONITOR)
    EI ; enable interrupts after IN execution
    IN A,($FF) ; back to MONITOR page 

The same page selection can be also achieved by using IN A,(C) with B = page number and C = $FF.

Writing a page is easy with nvRAM as support. Since it is like an SRAM, you only need to swap HL and DE.

For a NOR-Flash, it is a little more complex as you need first erase a sector through a sequence of bytes with specific addresses. And then you can program each byte in the sector with another sequence of bytes with specific addresses. I won't detail them for the moment because emulating a nvRAM extension is enough for me.

Would it be possible to add this emulation for a 128/256/512KB nvRAM? knowing that you may need a toggle to switch between the original ROM and the nvRAM (page 0 is a patched MONITOR)?

My purpose is to develop some tools and a file system to store some binaries in the nvRAM which are launchable through an IPL running at boot.

Or selectable theme.

It is not necessary to be like a SPA app on mobile.

The button state is not depending directly to the visibility state of the keyboard.

The assemble list and the dump list in the right pane should offer a vertical scrollbar.
Even it can be scrolled by the mouse wheel, but might be unavailable for mobile users.

ADC A,s and ADC A,n always return either 0 or 1 (value depends on the intended result being even or odd). Effectively, the instruction clears all bits except bit 0 when storing the result of the addition in the accumulator.

Appears to be a parenthesis bug in register.ts, line 384:

const q = (this.getA() + n + (this.getF()) & Z80_Register.C_FLAG);

Should be:

const q = this.getA() + n + (this.getF() & Z80_Register.C_FLAG);

v1.2.4 or later cannot be installed. Below is the log.

...
Running "browserify:build" (browserify) task

Error: Cannot find module 'js-cookie' from '/Users/hogehoge/.nvm/versions/node/v12.13.0/lib/node_modules/mz700-js/lib'
Warning: Error running grunt-browserify. Use --force to continue.

Aborted due to warnings.
npm ERR! code ELIFECYCLE
npm ERR! errno 6
npm ERR! [email protected] postinstall: grunt release
npm ERR! Exit status 6
npm ERR!
npm ERR! Failed at the [email protected] postinstall script.
npm ERR! This is probably not a problem with npm. There is likely additional logging output above.

npm ERR! A complete log of this run can be found in:
npm ERR! /Users/hogehoge/.npm/_logs/2019-11-15T12_53_09_792Z-debug.log

The address textbox is not updated when the button is pressed or scroll.

The timer is not needed.
It's enough only to show the element.

After the width is change.

Because the letter cases of the module file name in the source code do not match to the names on the Linux file system.