[gen] Re: FW: Generator 0.3.1

James Ponder james at squish.net
Wed Aug 15 21:46:48 BST 2001


On Wed, Aug 15, 2001 at 05:41:10PM +0100, Richard Bannister wrote:
>  Message: I have to agree with Lord Trelmayne and srsauer from
> www.versiontracker.com on some level. To my knowledge, none of those
> problems were there with Generator 0.2.3 (else then the need to switch at
> times from accurate to fast rendering or vice versa, but I don't mind that).

What is he referring to here?

> Batman2 (scrambled rendering),

I don't have this one.

> Strider 2 (error 3),

What's an error 3?  Well, this appears to be the same problem as sonic 3d,
block stepping goes wrong (ends up at $212 BRA $212 - I get a blank screen)
but single stepping doesn't.  Also, seems to be a bug in the complex plotter
window code here - it's all corrupted although I can see the other layers below.

> Eternal Champions (black screen)

It dies with illegal instruction for me, although single stepping works :-)

> Ghost Busters (constant need to switch from fast to accurate
> and vice versa),

Huh. Neat.

> Golden Axe 2 (sound distortion)

This works ok for me, but I'll ensure to test this before next version incase
it's due to things I've removed for debugging.

> Land Stalker (distorted > rendering then black screen, fast renderer
>used to fix the issue on v0.2.3),

I don't have this one.

> Maximum Carnage (Generator freeze)...

It illegal instructions for me, works whilst single stepping (I notice a
pattern here!).

> Recomendations:
>  A quick menu to switch from faster to more accurate renderer or vice versa.
>  Modifiable Cheat database to keep the newly added codes so you don't have
> to add them every time you load the game afterward.

Well, accurate should never be worse than the faster one - that's just a
bug and I'm glad he's reporting it.

In answer to your question in another email, stepping can be different to
block stepping as interrupts occur in different places, sometimes code
in interrupt routines branch into RAM which may be setup slightly differently
if we delay the interrupt a few ticks - single stepping is completely
accurate as I've written the routines... block stepping isn't.  The problem
we're seeing is really strange, it's impossible to debug - I'm going to look
at the code now and see if there's an obvious bug somewhere.


Best wishes, James
-- 
James Ponder; www.squish.net; London, UK




More information about the gen mailing list