[Scummvm-devel] Dreamweb engine inclusion.

Eugene Sandulenko sev at scummvm.org
Tue Jun 14 14:01:33 CEST 2011


On 14 June 2011 14:45, Max Horn <max at quendi.de> wrote:
>> I do not think that there is a real sense of making that code readable
>> or something, but we can fix bugs in it as we do with other engines.
>>
>> Of course, a bit more brains could be added to the translator, such as
>> loop detection, but all in all, refactoring those 20k lines of
>> basically assembly code will bring us no gain.
>
> Really, no gain?
> So what if there are bugs in the dreamweb engine? Do we then go and fix the bug in the original assembler code, then run the regeneration process again?
Yes, no gain in refactoring all 20k lines of code.

Once we will make sure that we correctly translate original sources
into dreamgen.cpp, we will stop code regeneration.

And of course, the fixes should be applied to .cpp code, not to .asm.
This was used only for speeding up the tedious work of manual
instruction processing.

> fuzzie's email makes me believe that the current generated code triggers warnings.
Yes, there are warnings of some unused goto labels. Of course, those
are trivial to clean up.


Eugene




More information about the Scummvm-devel mailing list