dodleh wrote on 2021-01-03, 17:08:

Dosbox is, no doubt, a very powerful emulator with good performance and compatibility. However, I have noticed for a long time ( […]
Show full quote

Dosbox is, no doubt, a very powerful emulator with good performance and compatibility. However, I have noticed for a long time (more than 10 years) an issue with Dosbox (be it classic SVN or ECE).

I assume that Dosbox incorporates a mouse driver similar to cutemouse. I may be wrong but it seems that part of its behavior was seen before with cutemouse. For obvious reasons the Dosbox mouse driver is not a TSR, it is directly incorporated into the emulation backend, with no outside control. In general, it works quite right.

The problem: Microsoft Windows 3.x. I know and I fully agree that Dosbox is not made for this dos extender but it is clearly used as such. Graphics is fine, accuracy is good. On the other side, mouse handling is less than great. For instance, mouse sensitivity control from the Dosbox config file has no influence. Even worse, opening a dos-prompt window may crash Dosbox. However, even that is solvable and switching from full-screen to window mode can work. Now, the real issue: mouse movement is choppy (presumably due to interrupt handling but the mouse driver can also be blamed), and, worse, if a windowed dos prompt session is started, as soon as the application being run (ms-dos edit) uses the mouse, the entire Windows GUI crashes to dos.

I am pretty sure that there could be a way to disable the integrated mouse support in Dosbox so that a typical MS mouse driver can be loaded as a TSR. In that circumstance, running Windows 3.x might be a different experience (it has to be tested). Of course, a simple parameter to switch in the config file is the perfect approach but I have not seen any such implementation upto now.

I appreciate any discussion regarding this issue. Thank you.

Have a look at this Re: Mouse problem

The existing "mouse driver" is basically just a stub put there by the emulated BIOS, loading any other mouse driver will replace it. But having said that I haven't seen any of the behaviour with windows 3.1 that you describe.

Win3 uses DOSBox's emulation of the BIOS PS/2 mouse interface; so the driver, internal or otherwise, is not involved. Perhaps that's only the default when Win3 is installed in DOSBox; not sure if other options can be selected.

I highly appreciate all your responses and thank you for your deep insight!

I tested mouseoff and it works perfectly! Absolutely no problems anymore with handling the mouse in a dos-prompt box inside Windows 3.x after loading microsoft mouse driver (the TSR). My mistake, I haven't searched more in the forum, I appologize.

I was wrong about the Windows 3.x behaviour behind the scenes, thanks for corrections.

I remembered why PIF settings such as disabling "Emulate Text Mode" and enabling "Retain Video Memory" are needed (it was not for the mouse, but for another DosBox glitch). If you do not use these settings, when you start a windowed dos-prompt and then an app, or just a dos-prompt, when the command.com interpreter instance is the one supplied with DOSBOX and not the MS-DOS one, the dos window background turns green and it is garbled!

Ok, I would like to also share with you a super-smooth mouse.ini for Windows 3.x that I devised, tested extensively and seems to be very good!

Also, I would like to share some win.ini settings (nothing fancy here, just standard acceleration) that seem to be ok for Dosbox SVN derivatives:
[windows]
MouseSpeed=1
MouseThreshold1=2
MouseThreshold2=0

The mouse.ini file has to overwrite the one available in your windows 3.x installation directory. Have fun!