#Cspect: SpecNext Emulator für Windows, Linux and OSX by Mike Dailly - New Version

[Seven.FFF]

It’s definitely possible that it’s AVG. A while back my CSpect was randomly termining about a minute or two after launch, with no indication why in CSpect or Windows Defender.

As soon as I set up a Defender blanket folder exclusion on the parent folder of all my CSpect instances, it stopped happening, and has never happened since.

It was the Defender realtime scan. A/V programs really don’t like or play well with apps that open image files for writing without doing any locking.

Also check cspect.log in the root of the CSpect directory (it works best if your CSpect doesn't live in Program Files or Program Files (x86), as writes to those directories go to a shadow directory).

And check your Windows Event Viewer in the Windows Logs >> Application section. Any CSpect and .NET Framework unhandled exceptions will get an event log written there with Level = Error.

[Pobulous]

A lot of AVs use sandboxing (basically virtualizing the application to intercept and examine all actions) of unknown executables, and that can sometimes prevent the applications from running properly.

[TMD2003]

It’s definitely possible that it’s AVG. A while back my CSpect was randomly termining about a minute or two after launch, with no indication why in CSpect or Windows Defender.
As soon as I set up a Defender blanket folder exclusion on the parent folder of all my CSpect instances, it stopped happening, and has never happened since.
It was the Defender realtime scan. A/V programs really don’t like or play well with apps that open image files for writing without doing any locking.

Windows Defender is telling me it's turned off. I've manually added CSpect to AVG's "Allowed Apps" list. Still there is no change in CSpect's ability to launch.

Also check cspect.log in the root of the CSpect directory (it works best if your CSpect doesn't live in Program Files or Program Files (x86), as writes to those directories go to a shadow directory).

The last write tot he log is still 5th January, and I keep the folder on the desktop, so it's not a problem with "thou shalt not write to the Program Files folder".

And check your Windows Event Viewer in the Windows Logs >> Application section. Any CSpect and .NET Framework unhandled exceptions will get an event log written there with Level = Error.

There's nothing there that looks anything remotely like CSpect or .NET put it there.

A lot of AVs use sandboxing (basically virtualizing the application to intercept and examine all actions) of unknown executables, and that can sometimes prevent the applications from running properly.

So it's odd that it hasn't done this - plenty of Spectrum emulators I've tried before (Es.pectrum and SpecEmu would be excellent examples, and the sites they're downloaded from are considered "dodgy" by AVG) have been dumped in the sandbox, assessed and passed fit, even if it took a few hours. CSpect has never seen the sandbox, at all.

And before anyone asks, I did update .NET to version 7 before dropping the CSpect folder on the desktop in the first place, knowing this might be one reason that it would refuse to run.

[Seven.FFF]

And before anyone asks, I did update .NET to version 7 before dropping the CSpect folder on the desktop in the first place, knowing this might be one reason that it would refuse to run.

CSpect actually uses Framework 4.5.2, so you could try installing this runtime version explicitly.

FWIW there are two different streams of .NET. 1.0,2.0, 3.5 and 4.X are .NET Framework, which has now ceased development on (I think 4.8.2).

2.1, 3.1, 5, 6 and 7 are .NET Core, which is a complete multiplatform rewrite. (They skipped version 3.5 and 4 to try and avoid confusion between the two streams, although both streams had some 1.X and 2.X version numbers in common.)

[Seven.FFF]

Windows Defender is telling me it's turned off. I've manually added CSpect to AVG's "Allowed Apps" list. Still there is no change in CSpect's ability to launch.

I just remembered that it was Webroot silently killing CSpect, not Windows Defender. And it did this whether or not an exclusion was added. I stopped using it and moved to Defender, where exclusions work as expected. The point being, sometimes the only way to really know A/V software is not interfering with something, is to completely uninstall it and reboot several times. I don't think you can uninstall Defender unfortunately, but you could temporarily uninstall AVG to rule it out.

Another thing that I've only just thought of, is to try CSpect with -tv added to the command line. Some graphics cards are apparently not compatible with the shader he uses, and -tv turns off the shader.

[TMD2003]

Another thing that I've only just thought of, is to try CSpect with -tv added to the command line. Some graphics cards are apparently not compatible with the shader he uses, and -tv turns off the shader.

That didn't cure the problem either. And as far as I know I haven't got Webroot, as I've never heard of it, unless it's something that comes packaged with Windows 7.

I suppose at this stage I've always got the Linux version working if I need to use CSpect on the laptop.

[Seven.FFF]

Annoyingly, it's a blank file with no text in it.

Seems like you have a bad or partial CSpect download. My copy of beast.bat contains this:

Code: Select all

CSpect.exe -fullscreen -sound -w2 -60 -vsync -zxnext -mmc=./ beast.nex

I would definitely try to download latest CSpect again, ideally with A/V disabled during the download and the unzip.

[TMD2003]

I've just found that BEAST.BAT is blank on the main PC as well, so I must have accidentally erased it before copying the files over. It isn't a dodgy download.

ALSO: can CSpect be set up with a virtual printer, and by that I mean some kind of text output, the way Fuse can? I've been looking for it in all the READMEs, and there's nothing mentioned. It would be useful, when I'm comparing lists of values, to be able to LPRINT them to a text file, or LLIST a program so I can read it outside the CSpect window.

[Seven.FFF]

If you're just running batch files and not seeing CSpect launch at all, is CSpect.exe even there? It's extremely common for A/V to silently remove the exe. What happens when you double-click on CSpect.exe itself?

If the exe is there, try enabling fusion logging and see what is logged: https://stackoverflow.com/questions/255 ... ion-in-net

Running a Process Monitor capture while cspect.exe is launched might tell you something too: https://learn.microsoft.com/en-us/sysin ... ds/procmon

[Seven.FFF]

ALSO: can CSpect be set up with a virtual printer, and by that I mean some kind of text output, the way Fuse can? I've been looking for it in all the READMEs, and there's nothing mentioned. It would be useful, when I'm comparing lists of values, to be able to LPRINT them to a text file, or LLIST a program so I can read it outside the CSpect window.

The Next itself doesn't have a built-in printer port, so I guess there's nothing really to emulate. It does have a general driver mechanism, where people can write their own printer drivers for specific hardware. There's at least one driver released for wifi text printing (JetDirect style to tcp port 9100) but of course that needs a Next with an ESP-01 wifi module. Possibly CSpect's limited wifi emulation would work with that, but if not, it's possible to hook up a real hardware ESP-01 to CSpect with a USB adaptor.

There are some dev-related text file logging features in the CSpect debugger, reflecting the dev focus of the emulator.

NextBASIC has built-in capabilities to output streams to a file. See chapter 21 of the Next manual, and in particular the "o>filespec" channel. They should work just as well in CSpect as they do on the real hardware.

[TMD2003]

If you're just running batch files and not seeing CSpect launch at all, is CSpect.exe even there? It's extremely common for A/V to silently remove the exe. What happens when you double-click on CSpect.exe itself?

Nothing. It's there all right, it comes up as a 420 kB .exe file, but unlike on the main PC, it doesn't launch anything. The main PC will launch CSpect in -w2 mode as a 48K Spectrum is I just use the .exe file and not the .bat with all the required switches.

If the exe is there, try enabling fusion logging and see what is logged: https://stackoverflow.com/questions/255 ... ion-in-net
Running a Process Monitor capture while cspect.exe is launched might tell you something too: https://learn.microsoft.com/en-us/sysin ... ds/procmon

That all sounds like it's going to descend into PC LOAD LETTER territory, and I'm going to leave that until tomorrow when I can concentrate a bit better.

The Next itself doesn't have a built-in printer port, so I guess there's nothing really to emulate.

NextBASIC has built-in capabilities to output streams to a file. See chapter 21 of the Next manual, and in particular the "o>filespec" channel. They should work just as well in CSpect as they do on the real hardware.

Streams! Argh. I never had anything to do with those on the Spectrum. They make more sense on the QL, but it's really just for printing and drawing in different windows. I haven't reached this part of the manual yet, and it looks like I'd have to leave that.

[TMD2003]

LATEST QUESTION:

Does anyone have a problem with CSpect where the repeating keyboard runs away with itself? It's particularly annoying when the key is DELETE and it erases an entire line I've just written. Switching to GAME mode seems to stop it, but because this is considerably less convenient to type with on a PC keyboard, I prefer BASIC mode unless I'm using graphics characters... whereupon the repeat will immediately start again and not stop.

I still haven't worked out what I press to make it do this, and if it's a bug, or an annoying feature.

Likewise, sometimes, BASIC mode will get stuck in a kind of shift-lock where the DELETE key starts printing ) instead of deleting a character (as it's mapping to shift+0). Again, I don't know what I've done to make this happen, and whether or not it's a bug.

[Luzie]

...
Likewise, sometimes, BASIC mode will get stuck in a kind of shift-lock where the DELETE key starts printing ) instead of deleting a character (as it's mapping to shift+0). Again, I don't know what I've done to make this happen, and whether or not it's a bug.

Well, I presume it´s a bug. Discovered it to under actual Version of CSpect v2.19.0 (I´m on Winx64). CSpect had keyboard-problems in older versions. I think you may try an other version of CSpect.exe to get rid of the problem. And than wait for a bugfixed version.

[Luzie]

@TMD2003

LATEST QUESTION:

Does anyone have a problem with CSpect where the repeating keyboard runs away with itself? It's particularly annoying when the key is DELETE and it erases an entire line I've just written. Switching to GAME mode seems to stop it, but because this is considerably less convenient to type with on a PC keyboard, I prefer BASIC mode unless I'm using graphics characters... whereupon the repeat will immediately start again and not stop.

I still haven't worked out what I press to make it do this, and if it's a bug, or an annoying feature.

Maybe this fixes your problem?!:

New version as seen on: https://www.speccy.cz/#newest/all

From: http://dailly.blogspot.com/2023/01/cspect-21903.html

Wednesday, January 18, 2023
#CSpect 2.19.0.3

#CSpect V2.19.0.3 changes

Massive update, with virtually all rendering components touched to properly allow for blending modes. While there has been a concerted effort to beta test this, please report any rendering oddities that you see on the Spectrum Next discord channel.
Also in this version, is a major update to audio/video syncing, producing far smoother and more playable results. The recommended command line for gamers will now be "-60 -vsync [game]". This should give full audio, and nice smooth gameplay.
General timing has also been improved, so without audio running at 60hz and vsync in fullscreen, everything should "glide" nicely again - as it used to in the C++ version.

Plugin authors be aware of a change to the Plugin system. Due to the new Audio/Visual syncing system, the emulator "could" now be running on a different thread, and as such, you can no longer do GUI/OS work from callbacks. ALL UI/OS work must now be managed via the new OSTick() callback. Like the Tick() callback, but this is always called from the OS thread, allowing you to open windows, and refresh/draw them. Please see updated plugins on my github for further example code.

Hope you all enjoy this one, it's been some time in the making - thanks to all who helped beta test it.


V2.19.0.3

Fixed Timex Hires non-ULANext ink/paper orders
Fixed Timex Hires ULANext ink/paper orders
Added wildcard support in F_OPENDIR ($a3)
Added better exception reporting to the LOG for loading plugins
Fixed DeZog plugin building - DLL has been properly updated with new Plugin interface


V2.19.0.2

Fixed up keyboard cross threading issues
Fixed up File open cross threading issue
Added a new OSTick() call to plugins to allow UI/OS function calls - please note the changes in plugins to avoid crashes
Fixed Timex hires with ULA Next ink/paper shift mode
Added a "Standard Plugin Keys" section to the readme.txt file


V2.19.0.1

Fixed 2 profiler crashes, when clicking on the profile window in different states
Fixed ULA last line drawing
Fixed sprite in border bugs with new screen rendering
Fixed L2 640x256 and 320x256 clipping
Fixed Timex border colour issues
New Audio/Video syncing system
Fixed 320x256 window


V2.19.0.0

Fixed a crash in Mode 6 and 7 when the tile window was larger than the screen
Fixed blending of tiles to border when in Mode 6 and 7
Fixed border when in Mode 6 and 7 when ULA is being used in blending
Fixed an esxDOS emulator issue, when a game requests too many bytes than is left in the file, and it fills the extra space with 0s resulting in game crashes
Timing fixed. 50hz and 60hz were both running slowly/weirdly both with and without sound. Should now be correct.
-60 and -VSYNC should now be perfectly smooth again
-fps added to show FPS on title bar
-freerun disable all timers and run as fast as we can (must use -sound as well)
ULA Y scrcolling fixed to use Yscroll value MOD 192 (instead of anything over 192 being 0)
ULA+TILE Stencil mode added. Stencil will use any ULA mode (normal, timex, lowres)
Mode6/7 blending can now be used with stencil mode (where ever the hardware allowed it)

Download: #CSpect V2.19.0.3 Emulator: http://www.javalemmings.com/public/zxne ... 19_0_3.zip

[Luzie]

btw.: RTC-PlugIn updated too for the latest #Cspect v2.19.x : https://github.com/Threetwosevensixseve ... iki/RTCSys

[TMD2003]

I've just noticed CSpect is nagging me about a new version. I will grab it immediately and see if the same issue occurs. Presumably what I mentioned is what Mike means by "keyboard cross-threading issues". It seems to be an issue with the EXTEND-plus-arrow features, and accidentally hitting shift and CAPS LOCK together, which is a common problem when you're me.

[Luzie]

Mike Dailly

https://dailly.blogspot.com/2023/01/cspect-21910.html

#CSpect V2.19.1.0 released.

Minor update, fixing some issues with the large rendering change, and adding super useful debugger command macros.

Rewrote command line processor
Added debugger macros to the command line '+def[0-9]"g $8000". use ";" for multiple commands: '+def[0-9]"g $8000;m$4000". ALT+[0-9] to play macro.
Fixed a crash when L2 left clip was > L2 right clip
L2 no longer renders with top clip was > L2 bottom clip
Fixed a bug in the ULA rendering where the border colour was being set to the fallback colour, and not the border colour (which matches global transparancy).
Updated TileTest screen mode test app
OpenTK upgraded to v3.3.3

Download: http://www.javalemmings.com/public/zxne ... 19_1_0.zip

[TMD2003]

Fixed a bug in the ULA rendering where the border colour was being set to the fallback colour, and not the border colour (which matches global transparancy).

That might just have boned the spectacularly brilliant piece of programming that you're all about to see... as soon as @Jbizzel reviews it. Because I'm Living After Midnight, though, I'll have to check if it works with this new CSpect in the (proper) morning.

[Luzie]

From: https://dailly.blogspot.com/2023/03/cspect-21920.html

Thursday, March 02, 2023
#CSpect 2.19.2.0

#CSpect V2.19.2.0 changes

Minor update, fixing some issues with the large rendering change, and adding super useful debugger command macros.

Added new debugger command "LOAD <filename>,<address>[,length]" where address can also be ""<bank>:<offset>""
Copper writes fixed up, wasn't executing all CPU TStates, so going slower than it should have been.
Updated screen drawing to be later in the line, to account for left border better.
Added a little Time base correction when audio/video buffers overflow instead of just throwing the spare frames away.
AY chips now respect the ABC,ACB and mono selection for each chip.
Added 2 new iCSpect.GetGlobal() calls; ""eGlobal.low_rom"" and ""eGlobal.high_rom"" to return the ROM/RAM at the low/high position
Added LoadFile() to iCSpect to load a file from SD card or relative file mmc path.

Download: #CSpect V2.19.2.0 Emulator: http://www.javalemmings.com/public/zxne ... 19_2_0.zip

[Luzie]

https://dailly.blogspot.com/2023/03/cspect-v21921.html

Sunday, March 05, 2023
#CSpect V2.19.2.1

#CSpect V2.19.2.1 changes

Minor fix for the debugger screen, and a couple of quick additions.

Fixed the debugger screen so it's no longer transparent
.NEX file start delay added to loading of NEX files
-nodelay added to skip a .NEX file start delay
-rot90 Rotate the display 90 degrees
-rot180 Rotate the display 180 degrees
-rot270 Rotate the display 270 degrees
Command line processing cleaned up a bit
cspec_win.dat file format changed. Now includes a version [0], and screen rotation [1]

Download: #CSpect V2.19.2.1 Emulator: http://www.javalemmings.com/public/zxne ... 19_2_1.zip

[Luzie]

From: https://dailly.blogspot.com/2023/05/cspect-v21930.html

Sunday, May 14, 2023
#CSpect V2.19.3.0

#CSpect V2.19.3.0 changes

Some debugger and Layer 2 fixes, and fixing the streaming API in the esxDOS emulation system.

Extended the debugger's "Display" view to show the whole screen
You will now be prompted to install OpenAL if it can't be found (on windows)
Fixed Layer 2 pixels in the border area
Fixed the esxDOS emulation streaming, allowing Pogie to run with audio from the command line again
Added "INPORT <16bitport>" and "OUTPORT <16bitport>" breakpoints in the debugger, allowing a break on a read/write to a port
Added a new "TONE" command to the debugger. This switches on a single tone to the audio buffer to test if there are system playback issues

Download: #CSpect V2.19.3.0 Emulator : http://www.javalemmings.com/public/zxne ... 19_3_0.zip

[Luzie]

Von: https://dailly.blogspot.com/2023/06/cspect-v21943.html

Friday, June 09, 2023
#CSpect V2.19.4.3

#CSpect V2.19.4.3 changes

Fixing up some audio issues, and a crash.

Reverted the ZIP file detection to fix SDCard access.
Updated some demos to fix keyboard reading equates

Download: #CSpect V2.19.4.3 Emulator: https://www.dropbox.com/s/6hcl37zyqqars ... 3.zip?dl=1

[Luzie]

From: https://dailly.blogspot.com/2023/07/cspect-v21944.html

Saturday, July 22, 2023
#CSpect V2.19.4.4

SNasm V3.2.1.0 changes

Some updates and fixes to both CSpect and SNasm

Equates can no longer start with a local label symbol
Locals are no longer reset by Equates/Sets/Rb/Macro etc.
Fixed LDDRX instruction byte codes
Proper Version added to assembly EXE
Change internal LEX processing and state saving/stacking

#CSpect V2.19.4.4 changes

Some updates and fixes to both CSpect and SNasm

OUT ($ff),$40 will not disable ULA Vblanks, while OUT ($ff),$00 will now enable them (see NextReg $22)
New debugger command "NEXTBRK" will stop in the debugger after a Next instruction has been executed

Download: #CSpect V2.19.4.4 Emulator: https://www.dropbox.com/s/vbi5joh78zvwa ... 4.zip?dl=1

[Seven.FFF]

CSpect Plugins v1.5 is released. These are my plugins, not Mike's (which confusingly have the same github repo name).

Added Pi UART support to UARTReplacement plugin.
Added support for mapping Pi GPIO pins 4 and 5 to serial DTR and RTS.
Fixed UART RX slowdown caused by sleeping 1ms every time the RX buffer is emptied.
Changed UART RX buffer from a polling- to an event-driven model.
Updated all plugins to work with CSpect 2.19.4.4.

https://github.com/Threetwosevensixseve ... ns/release

[PeterJ]

#CSpect V2.19.5.0 Released

https://dailly.blogspot.com/2023/12/cspect-v21950.html


Lots of fixes to #CSpect this time.
Fixed Layer 2, 320x256 right clipping
Fixed a sprite clip bug that was causing a hard crash
Optimised sprite rendering a little
Adding better streaming support through esxDOS
Adding in better config mode support
Copper can now generate IRQs properly
Fixed stackless NMIs
Fixed some DIVMMC RAM/ROM stuff
Fixed some DIVMMC address paging
Fixed a debugger memory window crash
First pass at a sprite viewer
Fixed "EQUs" being used as debugger address symbols
Added NextReg $B2 for extended MD Pad buttons
Fixed DMA continuous mode when also in Prescaler mode - though, please don't do this! Note: This is a total fudge, it may break.
Fixed DMA status register. DMA End flag, and 1 byte transferred flag now works.