Can you determine more precisely the moment it broke ?
If you are on windows you can download a lot of builds from here
http://builds.tharos-online.de/

It doesn't need to be the exact date, but a few weeks ago is too vague considering a lot of has happened the last few months actually

The problem is that I cannot be more specific. You see, I had not run win31 in dosbox for maybe 2 weeks. But I can remember that I did not have these problems then.

As I have said before: This problem is not related to one program. Sometimes, I open a program, and the first time I want to access the menu bar, it will hang (precisely: the mouse is not responding any more). I then have to press ctrl+F10 to give the control over the mouse back to Linux.

On the other hand, hoe can I be 100% sure if it is a dosbox bug? It MAY be that I have somohow manage to corrupt my win31 ... . But actually , I don't think so. So, I think if I'm the only one with these problems ... then I just have to reinstall win31 in dosbox. BUT, if there are more people with similar problems, then it is a dosbox bug.

in my experience it's easy to corrupt windows. I had it happening a few times at least.

The only way that you can be sure that it is a dosbox bug is that when it works in one version of DOSbox and crashes in antoher version of DOSBox. (and even then it is not 100 % foolproof)

As I do not have another build at hand: The only thing I can say for sure is that these problems do not appear with ver 0.65, which I still have installed as a kind of fallback. So I guess that I have not corrupted my win31 installation.

If you could compare 0.65 to CVS with just normal Windows applications and document when this instability happens then it would be very helpful.

CVS: I open the Main-ProgramGroup. Then I start the File Manager. Then I close the File Manager. Then I try to close the Main-ProgramGroup-window. I click into the menu, the menu appears, but it is blank (!!!). The mouse stops responding. I have to exit dosbox, releasing the mouse before.

0.65: No problems.

settings in cvs are ems=true + core=dynamic ?
0.65 I assume ems=false + core=normal

Yes, these are the settings.
Actually I do not know if dosbox switches to the dynamic core in the cvs-build, because my settings are: core=auto; cycles=auto. The cycles change to Max as soon as I start windows.

It does not matter if I set ems=true or ems=false in the cvs-build.

Try with fixed cycles, like cycles=30000 or so.

I could try, but: This is a new problem. A few weeks ago it did not matter if the cycles were max,auto,10000 or 30000.

I have tried to set the cycles to a fixed value (I tried out cycles=12000). The problems remain.

There are no problems with 0.65, so it cannot be that I have corrupted my win31 installation (is this a logical conclusion?), right? 🤣 🤣

You could try forcing the core in CVS

core=normal
core=full
core=dynamic

Try each seperately and report back.

Ok, dosfreak, the results are like this:

core=normal : NO problems. Windows acts like it did a few weeks ago (a bit slower, naturally !!)

core=full: Does not work at all. Windows is not even able to open the progman-window

core=dynamic : The same problems I have reported in my previous posts.

I hope the information is precise enough now.

Is that a straight cvs build?

yes, I have compiled it directly from the official sources.

why don't you first try the 0.65 release to make sure you have not corrupted your windows 3.x?

he allready did

sorry, I overread this.
Edit: just built a recent CVS and will try with Windows 3.x
Edit2: hmm, works without a problem for me

I have also built the cvs-sources again (10 minutes ago) to assure the problem has not been due to some weird compilation error.

But: The problems remain. core=normal works, whereas core=dynamic or core=auto cause weird stability problems.