BlackBox Framework Center

I think we should create a new issue.
But first please explain what the issue is all about.

- Josef

I have prepared a fix as proposed by luowy. Here is the link to the diffs: https://redmine.blackboxframework.org/projects/blackbox/repository/diff?utf8=%E2%9C%93&rev=aa5c4e96babc7fc97c92d9894b2d9986e919269c&rev_to=e2e743583046446ef1df515d8f017783f2387960 @luowy: I have added assertfn to the ...

I do agree that ASSERT(FALSE) should not be reported by the compiler but at runtime. If this is in line with Oberon-07 it is a good sign that it is the right choice and it eliminates a difference. The compile time evaluation of asserts is driven by the desire to catch as many errors as possible at c...

looks good to me.

Can there be any reason why those shortcuts were not supported originally?
I mean, for example, conflicts with shortcuts used by Windows or similar aspects.

- Josef

luowy, thanks for the explanation of the a1/a2 cache mechanism for the AX register value. I didn't know about that so far. I have pushed the changes to branch issue-#207. The diff is here: https://redmine.blackboxframework.org/projects/blackbox/repository/diff?utf8=%E2%9C%93&rev=96b3e9ff06545996...

@luowy, thanks for he info that we are in sync now. There is one more subtle detail to look at: It is the line marked <--- below. I don't know what a1 and a2 are used for and if that is of any relevance for our cases. It seems to be of relevance in case a comparison operation is converted into a boo...

@luowy here is the decoding I get from your example: 00000001H: 55 push ebp 00000002H: 8B EC mov ebp, esp 00000004H: 83 C4 FC add esp, -4 b := (b OR ~f) & f; (*trap*) 00000007H: 80 7D FF 00 cmp -1[ebp], 0 0000000BH: 0F 85 00 00 00 00 jnz 0 (00000011H) 00000011H: C6 45 FF 00 mov -1[ebp], 0 b := (...

maybe using assertions in SetPage is too restrictive.
Falling back to a useful default may be the alternative.

In any case, I would like to know where the strange settings come from.
All we need for finding that out is a user that is willing to help us diagnose the problem.

- Josef

@luowy, there is no trap in my solution either. There is a subtle detail with a missing Free(x), though. I mean, there is a register allocated at the beginning but in my proposal there is no Free(x) before changing x into a constant. Please check. The difference is that I am returning a Con item, no...

@luowy It would actually be sufficient to convert the conditional expression into a boolean constant for such cases, I think. As a side effect, it produces more compact code, however, in such exotic cases this is a minor point. My tests are successful, at least. DevCPC486.LoadCond IF y.offset >= 0 T...