LisaList2

Advanced search  

News:

2022.06.03 added links to LisaList1 and LisaFAQ to the General Category

Pages: 1 ... 3 4 [5] 6 7 ... 12   Go Down

Author Topic: LisaEm 1.2.7-RC4 support bug reports  (Read 81874 times)

rayarachelian

  • Administrator
  • Hero Member
  • *****
  • Karma: +105/-0
  • Offline Offline
  • Posts: 772
  • writing the code,writing the code,writing the code
    • LisaEm
Re: LisaEm 1.2.7-RC1 support bug reports
« Reply #60 on: August 04, 2020, 08:01:56 pm »
I've pushed the RC2 source code up at github, I couldn't yet push the macos x binaries.
Logged
You don't know what it's like, you don't have a clue, if you did you'd find yourselves doing the same thing, too, Writing the code, Writing the code

D.Finni

  • Sr. Member
  • ****
  • Karma: +38/-0
  • Offline Offline
  • Posts: 135
Re: LisaEm 1.2.7-RC1 support bug reports
« Reply #61 on: August 06, 2020, 12:11:11 pm »
Quote from: rayarachelian on August 04, 2020, 08:01:56 pm
I couldn't yet push the macos x binaries.
I couldn't wait.  ;D I back-ported your Generator changes to an older version of LisaEm and compiled it on my Intel Mac mini with OS X 10.5.

Suffice it to say, your addressing mode fix fixed a lot more than just those two or three main problems. There are several other little things that are now working correctly too.

You and this story should be on Slashdot.
Logged

stepleton

  • Sr. Member
  • ****
  • Karma: +127/-0
  • Offline Offline
  • Posts: 424
Re: LisaEm 1.2.7-RC1 support bug reports
« Reply #62 on: August 06, 2020, 01:39:39 pm »
I've compiled my own copy on my Linux laptop. Things went pretty well during building (I changed things to put the wxWidgets build in a subdirectory of my local repo---I am happy to work out of just this one directory and don't expect to need multiple LisaEms), but the screen updates at about a 1Hz refresh rate (regardless of the menu setting in Display). The Lisa mouse pointer does not appear to track the system mouse well---it positions itself halfway between the screen's top-left corner and the system mouse pointer location in the emulator window. Has anyone encountered this behaviour before? I hope I'm not missing a FAQ...
« Last Edit: August 06, 2020, 01:49:38 pm by stepleton »
Logged

rayarachelian

  • Administrator
  • Hero Member
  • *****
  • Karma: +105/-0
  • Offline Offline
  • Posts: 772
  • writing the code,writing the code,writing the code
    • LisaEm
Re: LisaEm 1.2.7-RC1 support bug reports
« Reply #63 on: August 06, 2020, 02:53:44 pm »
Quote from: D.Finni on August 06, 2020, 12:11:11 pm
Quote from: rayarachelian on August 04, 2020, 08:01:56 pm
I couldn't yet push the macos x binaries.
I couldn't wait.  ;D I back-ported your Generator changes to an older version of LisaEm and compiled it on my Intel Mac mini with OS X 10.5.

Suffice it to say, your addressing mode fix fixed a lot more than just those two or three main problems. There are several other little things that are now working correctly too.

You and this story should be on Slashdot.

Glad it helped. So here's what happened, I finished this code on Sunday night and yelled "Woo Hoo" quietly so as not to wake the kids and went to bed as it was nearing midnight.

Monday I finished cleaning up the code and started compiling macos binarinies and I have them on the  big laptop, but didn't finish vetting them so didn't push them up.

Tuesday around 2pm the hurricane hit and lost power, and I've been without power since. I managed to copy all the stuff off to the chromebook which has galliumos on it and can run a very long time before losing the battery on the big laptop but have no power at home other than charging off the car and no internet - just wifi hotspot off the cellular network. so not sure how long this chromebook will last. right now it says it's got 8hrs left, so hope to make good use of that. i have some solar 5v power banks that we've been using to charge our cell phones off, but that's it. PSE&G says they expect to fix power for 85% of the outages by tomorrow night, so that sucks. :(

luckily when I went on vacation a couple of years before i built a step up 5v to 19.5v transformer using COTS parts off amazon so I can use the phone charger to charge up the chromebook, but it drains them very quickly. right now I've got two of those outside charging from the sun and a couple more I plugged into a car battery so hope that will do until power comes back.

the 2nd biggest challenge now is that we have to finish off all the fresh food in the fridge b4 it goes bad. luckily the house has a gas stove and water heater so we can at least cook and shower. last night was very rough, no sleep bc it was 95f in the bedroom and the noise from the neighborhood generators was also painful. was nice and cool this am so managed to get some more sleep.

I guess another pain to add to the list of how much 2020 sucks, right? we've been through these things before, and did have a lot of UPSs, but ofc they have limited run time.

the bug itself is one i introduced bc when I added support for those address modes to generator back in 2006. I forgot to add source vs destination and felt pretty lame when I saw that's what it was. so it was confusing the source operand with the destination for this move.l I posted about earlier. so that's a 13 year old bug. :)

If you see other stuff it fixed that you recall, pls report it here so I don't wind up chasing stuff that's fixed later on.

I'm on a cell phone when writing this, so it's hard for me to reply, I had to dig up the pass for my lisalist2 account so wasn't fun.

and thx for your kind words.
Logged
You don't know what it's like, you don't have a clue, if you did you'd find yourselves doing the same thing, too, Writing the code, Writing the code

rayarachelian

  • Administrator
  • Hero Member
  • *****
  • Karma: +105/-0
  • Offline Offline
  • Posts: 772
  • writing the code,writing the code,writing the code
    • LisaEm
Re: LisaEm 1.2.7-RC1 support bug reports
« Reply #64 on: August 06, 2020, 03:03:21 pm »
Quote from: stepleton on August 06, 2020, 01:39:39 pm
I've compiled my own copy on my Linux laptop. Things went pretty well during building (I changed things to put the wxWidgets build in a subdirectory of my local repo---I am happy to work out of just this one directory and don't expect to need multiple LisaEms), but the screen updates at about a 1Hz refresh rate (regardless of the menu setting in Display). The Lisa mouse pointer does not appear to track the system mouse well---it positions itself halfway between the screen's top-left corner and the system mouse pointer location in the emulator window. Has anyone encountered this behaviour before? I hope I'm not missing a FAQ...

Sorry Tom, I'm sure I introduced this the last time I tried to fix macos x off the top of my head this is all in lisaem_wx.cpp, if you diff the last last two RCs when it was working vs now you should see where the changes to the wxui were. lkely in the onpaint_skin and onpaint_skinless methods. possibly a change in the type of bitmap for my_lisabitmap and related.

there were other changes around mouse mapping with zoom perhaps being that the big laptop has a 4k display the mapping between the mouse and display and zoom levels is the issue. I think I put in some tables for each OS that help scale for the display. it's likely that my 4k laptop running GTK behaves very differently than yours, I guess I'll notice it on gallium next time I compile and hopefully will fix it. I don't know if that was the right thing to do or not but it seemed to work on the 3-4 physical machines I tried the last time as well as the macos vms I used.

You can try cheating by replacing lisaem_wxui.cpp from the one that work and see if that fixes it for you for now. but there's a couple of other bugs I fixed there - I think screenshot file naming that were previously broken in the one you last tried.

also wx3.1.4 is out, and I wanted to test against it too - not sure if it will help or cause more issues.
Logged
You don't know what it's like, you don't have a clue, if you did you'd find yourselves doing the same thing, too, Writing the code, Writing the code

rayarachelian

  • Administrator
  • Hero Member
  • *****
  • Karma: +105/-0
  • Offline Offline
  • Posts: 772
  • writing the code,writing the code,writing the code
    • LisaEm
Re: LisaEm 1.2.7-RC1 support bug reports
« Reply #65 on: August 06, 2020, 03:08:43 pm »
oh, one more thing I do remember changing some stuff around refreshing because previously it was refreshing constantly like hundreds of times per second, but I don't recall where that code was or what it was. I'll try to dig, that's possibly for the refresh rate. on macos x it was blitting a blank white screen with just the mouse when LOS moved the mouse around, plus some pixels underneath the mouse, and then alternating between that and the screen refresh, but I guess on GTK it was covering that up with a full repaint way too often. so when I fixed it on macosx I broke it on GTK.

Isn't it fun how even if you have a cross platform GUI framework you still need to be picky and careful for each OS?
that's another thing for the future, what to replace wxwidgets with? or if I'm doing it wrong, how do I do it right?
Logged
You don't know what it's like, you don't have a clue, if you did you'd find yourselves doing the same thing, too, Writing the code, Writing the code

stepleton

  • Sr. Member
  • ****
  • Karma: +127/-0
  • Offline Offline
  • Posts: 424
Re: LisaEm 1.2.7-RC1 support bug reports
« Reply #66 on: August 06, 2020, 06:40:59 pm »
oh no! Hope you get the power back soon. Thanks for the advice---will have to try some other time if you don't fix it first!
Logged

rayarachelian

  • Administrator
  • Hero Member
  • *****
  • Karma: +105/-0
  • Offline Offline
  • Posts: 772
  • writing the code,writing the code,writing the code
    • LisaEm
Re: LisaEm 1.2.7-RC1 support bug reports
« Reply #67 on: August 11, 2020, 09:11:37 pm »
« Last Edit: August 14, 2020, 12:49:38 pm by rayarachelian »
Logged
You don't know what it's like, you don't have a clue, if you did you'd find yourselves doing the same thing, too, Writing the code, Writing the code

rayarachelian

  • Administrator
  • Hero Member
  • *****
  • Karma: +105/-0
  • Offline Offline
  • Posts: 772
  • writing the code,writing the code,writing the code
    • LisaEm
Re: LisaEm 1.2.7-RC2 support bug reports
« Reply #68 on: August 16, 2020, 11:17:35 am »
About the video/updates/mouse issues Tom reported:

At first I added back the video HZ refresh display and saw it was way off, but this is because most of the time there isn't much to update on the display except when a dialog box, or window changes or the mouse moves. I also added a checkbox to force the display update at all times which flags all video RAM as dirty, forcing a full repaint each time.

At some point it jumped from something like 2Hz to several KHz, so that wasn't right either, and it doesn't seem to care what refresh rate you select from the menu, so there's a disconnect there.

Next, I've been playing around with adding a 2nd wxTimer for video updates, which isn't turning out too well either. I've tied this to the refresh rate in the display menu, so when you change between say, 60Hz to 20Hz, it sets the timer to fire that often and then call Refresh and Update. I've found two things, even though Refresh is called, it doesn't really create an OnPaint event as often, and worse, it's pinning the CPU so badly that it prevents the pull down menus (the usual File, Edit, Display, Throttle, etc.) from coming down. I also added menu options for 120Hz and 100Hz, but it's behaving badly even at 20Hz, consuming far more CPU than available, so these are unlikely to remain.

Code: [Select]
void LisaEmFrame::OnDisplayTimer(wxTimerEvent& event)
{
    paintcount++;

    if (force_display_refresh) videoramdirty=32768;

    Refresh(false,NULL);
    if (force_display_refresh) Update();
}

The original code uses a single timer to call the emulation loop and intersperses calls to Refresh/Update. So perhaps I'm going to return to this. The issue there is that the rate of this timer is tied to the CPU clock, so I'd have to figure out some gymnastics to keep the display refresh rate true.

On the other issue of the mouse, I added a checkbox to disable the or enable the mouse scaling. The big issue is that when using wx3.1.x+ GTK on a 4K display with GTK scale set to 2.0, getting the display resolution returns 1920x1080 while earlier versions of wx3.0.x return the physical resolution of 3840x2160. However, I found there is a GetPPI call that on the 4k display returns 256x256 - not sure this is correct and certainly it's incomplete as there doesn't seem to be a method to get the physical size of the display, so you can't tell the actual physical resolution. So not sure what would happen, on say a 56" 8K TV vs a 4k 17" monitor, etc. Alternatively I'd have to write some direct GTK code to ask GTK what the physical resolution is, which I'm not sure if it will return a scaled or actual version. Meh.

I'll move the code to an older linux laptop and a chromebook to test the mouse scaling and GetPPI() call there to see what it returns there and what the mouse scaling should be.
Logged
You don't know what it's like, you don't have a clue, if you did you'd find yourselves doing the same thing, too, Writing the code, Writing the code

rayarachelian

  • Administrator
  • Hero Member
  • *****
  • Karma: +105/-0
  • Offline Offline
  • Posts: 772
  • writing the code,writing the code,writing the code
    • LisaEm
Re: LisaEm 1.2.7-RC2 support bug reports
« Reply #69 on: August 18, 2020, 10:35:12 pm »
Looks like wx3.0.5 builds and works on macos x 10.5, so that should bring in some nice fixes.

Looks like GetPPI is ~130x166 on the chromebook and disabling the mouse scaling there works, so that's the fix for the GTK mouse issues.
Logged
You don't know what it's like, you don't have a clue, if you did you'd find yourselves doing the same thing, too, Writing the code, Writing the code

D.Finni

  • Sr. Member
  • ****
  • Karma: +38/-0
  • Offline Offline
  • Posts: 135
Re: LisaEm 1.2.7-RC2 support bug reports
« Reply #70 on: August 20, 2020, 12:16:54 pm »
I found crashes. I'm running Intel 10.5.8. There is a text file attached to this forum post with details of the crash.

Also the status line at bottom of window has a tendency to show multiple messages concatenated together.
Logged

rayarachelian

  • Administrator
  • Hero Member
  • *****
  • Karma: +105/-0
  • Offline Offline
  • Posts: 772
  • writing the code,writing the code,writing the code
    • LisaEm
Re: LisaEm 1.2.7-RC2 support bug reports
« Reply #71 on: August 20, 2020, 06:22:13 pm »
Interesting, James Denton reported the same kind of thing against 10.15: crash and concated status messages - turned out to be an issue with the XCode version in my VM. He was on 10.15.6, I was on a 10.15.0 in the VM with an older XCode toolchain.

I'm upgrading to wx3.0.5, and where possible 3.1.4 for the next RC3, so that might help. I'll give it a test on 10.5.8 before I publish it.
Code: [Select]

Exception Type:  EXC_BAD_ACCESS (SIGBUS)
Exception Codes: KERN_PROTECTION_FAILURE at 0x0000000000020006
Crashed Thread:  0

Thread 0 Crashed:
0   lisaem-i386-10.05-wx3.0.4        0x0009fb49 wxConfigBase::Read(wxString const&, wchar_t const*) const + 206425

Unfortunately since debug wasn't compiled in, the crash report is almost worthless. Almost, however there is a clue that it has to do something with the config file. As a work around you might try renaming the config file - both the main one and the Lisa specific one config and trying it again, there's a clean-lisaem-prefs-macosx.sh script under scripts which will delete them, but renaming them will allow you to recover if that doesn't fix it. Alternatively, you can try:

Code: [Select]
cd lisaem-1.2.7
./build.sh clean build --debug --tracelog

And see what errors/messages turn up before the crash.

Or just wait a few days for RC3 :) Speaking of, I redid the video refreshes again, previously it was relative to the CPU speed, so if you picked 60Hz, and were at 5MHz CPU speed, it would  refresh as needed upto 60Hz. Now it's relative to the host clock. There's new two checkboxes under the Display Menu, now, one to force repaints whether it's needed or not and another to disable the mouse scaling.

I might remove the force-repaint one, not sure if it's really useful, it certainly will eat a lot of host CPU cycles, which may also limit the maximum 68k CPU speed.
I've also added back the video refresh rate meter in the status bar. The "D" next to it indicates vidram was dirty on the last pass, but the rate is no longer the actual number of refreshes/second, but the attempted refreshes, so if you say 60Hz, it'll be close to 60Hz and vary +/- some timing delta based on how busy the machine is and whatever else is going on.

There's a new default preference that checks the PPI of the display as a proxy for testing hidpi/retina displays, if it's less than 200 and on GTK, it disables mouse scaling, which should fix the other issue Tom reported, of the mouse being way off, but it might be useful elsewhere. However, if you switch between hidpi and lower res displays, the setting will stick to whatever was saved in preferences...

I'm working in parallel on x/profile disk image support, so if you've taken a disk image of an X/ProFile CF card with dd you should be able to use it with a future LisaEm, but it won't be ready for RC3, though it'll likely be in the release 1.2.7.
« Last Edit: August 20, 2020, 06:54:39 pm by rayarachelian »
Logged
You don't know what it's like, you don't have a clue, if you did you'd find yourselves doing the same thing, too, Writing the code, Writing the code

D.Finni

  • Sr. Member
  • ****
  • Karma: +38/-0
  • Offline Offline
  • Posts: 135
Re: LisaEm 1.2.7-RC2 support bug reports
« Reply #72 on: August 20, 2020, 09:11:13 pm »
Quote from: rayarachelian on August 20, 2020, 06:22:13 pm
Unfortunately since debug wasn't compiled in, the crash report is almost worthless. Almost, however there is a clue that it has to do something with the config file. As a work around you might try renaming the config file - both the main one and the Lisa specific one config and trying it again, there's a clean-lisaem-prefs-macosx.sh script under scripts which will delete them,
Yeah, I thought this might be something to do with my old config file from version 1.2.6. I will wait for RC3.

The screenshot I posted is a separate problem. That dialog box won't go away even if I click OK. I had to force quit the application.

I had the mouse problem that Tom reported too.
Logged

rayarachelian

  • Administrator
  • Hero Member
  • *****
  • Karma: +105/-0
  • Offline Offline
  • Posts: 772
  • writing the code,writing the code,writing the code
    • LisaEm
Re: LisaEm 1.2.7-RC2 support bug reports
« Reply #73 on: August 20, 2020, 10:08:51 pm »
Quote from: D.Finni on August 20, 2020, 09:11:13 pm
I had the mouse problem that Tom reported too.

Is this on 10.5.8 or elsewhere?
Logged
You don't know what it's like, you don't have a clue, if you did you'd find yourselves doing the same thing, too, Writing the code, Writing the code

D.Finni

  • Sr. Member
  • ****
  • Karma: +38/-0
  • Offline Offline
  • Posts: 135
Re: LisaEm 1.2.7-RC2 support bug reports
« Reply #74 on: August 21, 2020, 11:12:26 am »
Quote from: rayarachelian on August 20, 2020, 10:08:51 pm
Quote from: D.Finni on August 20, 2020, 09:11:13 pm
I had the mouse problem that Tom reported too.

Is this on 10.5.8?
Yes, on 10.5.8.
Logged
Pages: 1 ... 3 4 [5] 6 7 ... 12   Go Up
« previous next »