Sierra's SCI System Scripts

[Collector]

If the Debug doc is 1988-03-31, then it is SCI0, circa the original KQ4. The one version of KQ4 I do not have.
I suspected it was SCI0 from the descrition of the debugging window. For SCI0 the debug switch should have been [Shift]+[Shift]+[the numpad '-']. Perhaps if this was written for the original SCI interpreter they carried over the same debug trigger from the AGI and changed it with subsequent versions. Wish I could find the original release of KQ4 to see if Alt+D triggers Debug.

[Kawa]

Interesting detail: this version of the terp has KRemapColors.

[troflip]

Sounds like it's at least late SCI1.1, then eh?

[Kawa]

If I take SCI.EXE or SCITESTR.EXE and run Catdate in 'em, I get some funky palette mistakes in the last few entries, around index 253 or so. Otherwise, it seems to run fine. So yeah, I could have something nice here.


Update: adjust 999.pal, reimport backgrounds... add a (StrSplit 1 253 64) in a tactical location... and you get this! That's not even the QfG4 Demo's RemapColors there, that's the full version from SCI2, in SCI11! (or rather, SCI16? 1.6? What?)

[Kawa]

They used Borland C++ and either TASM or MASM for the assembly files. I suspect they different versions for different tools.

This doesn't quite mesh with what the makefile tells me, incidentally. It references Microsoft by name but never Borland, there is no cl.exe (the program invoked by the makefile to compile) in my copy of Borland C++ (the one used for Wolf3D), and cl.exe seems to be (I can't check for certain right now) MS' compiler. And MASM is also MS.

Of course, most of this is half-asleep rambling on my phone, but... yeah.

[Collector]

I was noticing things that made me wonder if VS was involved in some way.

[Kawa]

It's so involved. I got an old copy of MSVC 1.52 and MASM, and except for two tiny errors I am not knowledgeable enough to correct, I can build all four targets.

These errors? In startasm.s, lines 104 and 135:

Code: [Select]

setjmp proc
...
longjmp procThe complaint? "Conflicting parameter definitions" compared to setjmp.i's line 16-17:

Code: [Select]

setjmp proto buf: jmp_buf
longjmp proto buf: jmp_buf, code: wordCommenting those two lines allows the targets to build, but they can't be run.

Edit: turns out the proper fix is to make the ones in startasm.s this:

Code: [Select]

setjmp proc buf: jmp_buf
...
longjmp proc buf: jmp_buf, code: wordThis allows them to build, but unfortunately, I got the same results.

Edit: if you wanna try it yourself, here's a handy minimal thing to put in your VM's search path. http://helmet.kafuka.org/sci/msvc15-minimal.zip

[lskovlun]

They used Borland C++ and either TASM or MASM for the assembly files. I suspect they different versions for different tools.

This doesn't quite mesh with what the makefile tells me, incidentally. It references Microsoft by name but never Borland, there is no cl.exe (the program invoked by the makefile to compile) in my copy of Borland C++ (the one used for Wolf3D), and cl.exe seems to be (I can't check for certain right now) MS' compiler. And MASM is also MS.

Of course, most of this is half-asleep rambling on my phone, but... yeah.

The message editor uses Turbo Vision.

[Kawa]

So it does. I was under the impression Mu meant the interpreter.

[Kawa]

Y'know what else SCI.EXE has that I haven't seen before in 256 color mode?

A built-in debugger.

I was trying to get a picture of an Oops error for stupid joke reasons, and decided to take the hard way. Instead of an oops error, I got a window, title "PMachine", content "'edgeHit' is not a selector for PseudoMouse.". Dismiss that, I got the classic SCI0 Debug window.

Awesome.

Fun fact, pressing V for Version reveals the following:

Version 1.001.097
Made on Aug 23 1993 at 12:03:27
in i:/interp/sci/ibm
by ChrisS
Release Version

...Please don't lie to me, Mr. Smith.

[MusicallyInspired]

Interesting. I haven't had a chance to test these tools yet, but does its debug mode work with most Sierra games?

[Kawa]

I haven't checked. I also haven't a full-size keyboard to trigger it with.

I expect it to work on most if not all SCI11 games, though. Besides the remapcolor thing... I suppose any attempt to use StrSplit might risk triggering the debugger?

[MusicallyInspired]

Heh. It does work. Debug and all.

[OmerMor]

Yeah, it does. 

It seems like Sierra was stripping the built-in debugger from production builds, while keeping it for development.

[MusicallyInspired]

I've been trying to get the editors to work but I keep getting "cannot be run locally" errors. I guess many of these were meant to be run on a closed network with other networked files or something. I "succeeded" once with PE256 and a SQ1VGA pic (that is, it display text that I could read instead of all black windows), but it kept popping up error messages about X and Y being too long or something. Interesting tools there for Messages and Message audio, though. Very interesting stuff!