VOGONS

Common searches


Search results

Display options

Re: 64-bit dynamic_x86 (patch)

For anyone testing 64-bit builds on windows, SDL can have a minor issue where it causes a crash in a message handler and the OS "helpfully" tries to permanently apply a compatibility fix to hide it. This was fixed in a later SDL-1.2 version and can be backported as a workaround: https://hg.libsdl. …

Re: 64-bit dynamic_x86 (patch)

GenReg::Clear() checks if DYNFLG_CHANGED is set. If so it calls Save(), which clears it. This isn't new code, it's copied verbatim from risc_x86.h. Maybe you should improve your CI workflow to build the most popular DOSBox configuration (32-bit + dyn_x86).

Re: 64-bit dynamic_x86 (patch)

Information for line 434 refers to the value of `dnew->genreg` after Clear (this pointer is set to 0 inside Clear if - I assume that's the case here - there's a cyclic dependency between dynrec and genrec (line 244)). Therefore whenever this code will start with (`dsynch->genreg == nullptr` and ` …

Re: 64-bit dynamic_x86 (patch)

The only static analysis issue that looked remotely valid to me was "Called C++ object pointer is null" for line 444 in risc_x64.h, but the analyzer has performed an invalid simulation (assumes dnew->genreg != dsynch->genreg but also assumes both values are NULL, making the first assumption invalid …

Re: Please add option to remove 'banner'

It's displayed for 600ms with the last 100ms fading out. That's blink-and-you-miss-it speed. And yet people still complain about it! When I was working on my wii build I never saw it either due to the tv mode changes taking too long to settle, so I replaced the wait with the virtual memory …

Re: Please add option to remove 'banner'

The logo was added because companies were selling DOSBox (as part of other products) without acknowledgement. So making it configurable would be counterproductive to its existence. This is a direct quote from the source code: /* Please leave the Splash screen stuff in working order in DOSBox. We …

Page 55 of 67