I've been trying to track down an intermittent heap corruption issue that occurs on our build machines when they run our full regressions suites.
I've built an executable with the madExcept 'crash on overrun' feature turned on and it happily runs in that mode on my laptop, and I can see the expected dramatic increase in memory use getting to the main menu. And I've since found some read overruns, but not writes overruns yet doing some simple tests on my laptop.
However, when I copy that same executable over to the regression machine where we run the full regressions in a RAM drive and launch it, it just shows the normal amount of memory use so I think the 'crash overrun' memory manager has not installed itself. Has anyone else ever seen that?
Any ideas how I might troubleshoot this or determine why the 'crash on overrun' checker isn't enabling?
Oh, just to offer some important context - this is using madExcept 5 with Delphi 10.1 Berlin in a 64-bit application since the intermittent AVs only happen in 64-bit builds of our application.
So I was very happy that madExcept 5 supports overrun checking in 64-bit applications now
I needed to copy madExcept32.dll and madExcept64.dll onto the Regressions machine and ensure they were in the PATH.
That's probably documented somewhere but I'd never read it.