SCI Companion V3 - alpha build notes/bugs/feature requests

[NewRisingSun]

I've lost sight of the various alpha versions flying around, but I'm using the SCICompanion.exe dated July 20th 2016. When recompiling the decompiled script 998 (original name "ACTOR.SC") of SQ1, the following sequence in Actor.setLoop:

Code: [Select]

(method (setLoop param1 &tmp theLooper)
(if
(= theLooper
(cond
((== argc 0) (super setLoop:) 0)
((not (IsObject param1)) (super setLoop: param1 &rest) 0)
((& (param1 -info-?) $8000) (param1 new:))
(else param1)
)
)
(if looper (looper dispose:))
((= looper theLooper) init: self &rest)is properly decompiled, but after recompilation, it becomes:

Code: [Select]

(method (setLoop param1 &tmp theLooper)
(if
(= theLooper
(cond
((== argc 0) 0)
((not (IsObject param1)) 0)
((& (param1 -info-?) $8000) (param1 new:))
(else param1)
)
)
(if looper (looper dispose:))
((= looper theLooper) init: self &rest)
)In other words,  the "super setLoop" in the first two conds somehow gets lost, which results in the wrong loops being displayed in SQ1's intro. Right now, I am working around this bug by rewriting the code as

Code: [Select]

(method (setLoop param1 &tmp theLooper)
(cond ((== argc 0) (super setLoop:) (= theLooper 0))
((not (IsObject param1)) (super setLoop: param1 &rest) (= theLooper 0))
((& (param1 -info-?) $8000) (= theLooper (param1 new:)))
(else (= theLooper param1))
)
(if theLooper
(if looper (looper dispose:))
((= looper theLooper) init: self &rest)
)
)
The other problem I had was that "Replace all" sometimes would never finish. The last issue is more of a small quibble: When recompiling an existing script that I opened by double-clicking on the resource in the Explorer, the selection in the Explorer is reset to the start after compiling. I understand that recompiling causes the list to be refreshed, but it would be great if the selection could stay on the resource on which it was before the list is refreshed.  Otherwise, SCI Companion works great for me.

If I may request an additional feature: the ability to produce a RESOURCE.TXT file from an existing set of RESOURCE.* packages, so I can use it with Sierra's MAKEVOLS to repackage a changed game with compression. This could be implemented as an option in the "Extract all resources" window. I have, more or less manually, written a file for SQ1 that MAKEVOLS will accept and yields identical RESOURCE.* files to the released game when fed with extracted unmodified resources.

[OmerMor]

The scummvm team found that SCI32 changed its handling of zero-offset exports: https://github.com/scummvm/scummvm/commit/55222ec

SV and SciCompanion are both affected by this change.
You can see this in Space Quest 6, where in rm500.sc sEndChapter1::changeState(6) is calling __proc64909_0 (originally ShakePlane) but the 64909 script has "export64909_0 = $0" (in SV) and the procedure is not detected and decompiled.


EDIT:
I have added this issue and the previous ones reported by NewRisingSun to the project's github page: https://github.com/icefallgames/SCICompanion/issues
I hope it'll help you track it.

[Collector]

He is the one who started VOGONS. The current style was originally designed by him and was recreated for phpBB3 by Qbix and myself. He has not been around in years and gave the board to Qbix. Curious that he never seemed to express much interest in the Sierra games at the time.

I know he started VOGONS.
I recall he joined scummvm to brush up his C++ skills.
He's working on SCI32 and making great progress; just see the commit history for the SCI engine and see how much of it is his work:
https://github.com/scummvm/scummvm/commits/master/engines/sci

Just noticed that snover has just registered here.

[OmerMor]

He is the one who started VOGONS. The current style was originally designed by him and was recreated for phpBB3 by Qbix and myself. He has not been around in years and gave the board to Qbix. Curious that he never seemed to express much interest in the Sierra games at the time.

I know he started VOGONS.
I recall he joined scummvm to brush up his C++ skills.
He's working on SCI32 and making great progress; just see the commit history for the SCI engine and see how much of it is his work:
https://github.com/scummvm/scummvm/commits/master/engines/sci

Just noticed that snover has just registered here.

Yes, I asked him to. I urged him to report his bugs with SCI Companion, but ended up reporting them before he got to do it himself.

[gumby]

Just noticed that the 'while' keyword isn't highlighted in the script editor as a keyword (like 'do').  Probably already fixed, but thought I'd point it out in case it got overlooked.

[Kawa]

So earlier today I was working on a new 2017 version of the Dating Pool demo, replacing some renders, the iconbar, and adding shadows to views and such. I found out that SCI Companion 3.1.0.3 doesn't compile things quite right. The iconbar's verb buttons wouldn't actually select their respective verbs (so you gotta right-click cycle instead) and the glint on the Firrhna Productions logo insisted on using loop 1 instead of 0, so it was a big ol' "DEMO" instead. Version 3.0.1.5 meanwhile, given exactly the same inputs, does not result in these issues.

Much like gumby said, this is probably already fixed, considering there's been a whole lotta commits since then, but I've no way to compile my own. Just thought it was an interesting thing to mention.

Fake edit: just downloaded 3.0.1.7 and recompiled, it started some time after.

Real edit: script 937 "IconI" has many differences. If anybody wants a copy...

Update edit: http://helmet.kafuka.org/sci/937.html -- some bnot are missing, and IconI::select and IconBar::select have large parts missing.

[Kawa]

Okay, so I compiled my own from the latest git. On the one hand, the compiler seems to be fixed. On the other, the thing crashes when I try to open a picture resource with an associated shp file, on Polygon.cpp line 507. Because on line 80, "it" is apparently empty so pValue is null.

[troflip]

Thanks for finding this (Beta tester ;-) )

Fixed:
https://github.com/icefallgames/SCICompanion/commit/99a2b38f462f54a0331b064c8a0e71f2756d9f90

[troflip]

I'm sure there will be other bugs,as I haven't done too much testing with SCI1.1 (which is why I haven't officially released it yet).

In particular, "Game properties"->"MAnage resources as patch files (good for source control)" doesn't work with SCI1.1.

Most of the new features involve SCI0 pics, but there is some new functinoality in views:
- you can shift a view's pixels up, down, left, right (I think SCI Studio even offered this!)
- you can rotate a view by any amount
- you can enable "onion skins" to see ghosts of the previous or next cel, to make animation easier

If you make your views in some external program though, then these aren't that useful.

You can "Extract all script strings" to make spell-checking easier.

[Kawa]

So I'm looking at a few SCI0 games in the latest SCI Companion and testing the EGA undither switch when...

owo what's this?


...oh!


And so, the downside of EGA undithering comes to light -- it doesn't work on manual dithers like the above and certain babe portraits in the Larry games?

I have a patch for this particular file to enhance the undither effect a bit, because I'm me and I am a bored me.

? not to mention their eyes and such, being sprites, are never undithered so that looks real weird in practice.

[lskovlun]

And so, the downside of EGA undithering comes to light -- it doesn't work on manual dithers like the above and certain babe portraits in the Larry games?

I don't think I've ever seen results this bad in FreeSCI, FWIW.

[troflip]

That Iceman pic isn't a problem with undithering per se, it looks like it just has a bug in it... they left the visual screen on when drawing the zig zag priority lines, and happened not to notice because the blue color lined up with the dither pattern (they also missed a fill spot behind the plant on the center left!). Not sure how FreeSCI would fix that...


SQ3 has some undither issues, like the scene with Roger's head and the window light shining on it.

[Kawa]

SQ3 has some undither issues, like the scene with Roger's head and the window light shining on it.

That'd be because the part you seem to refer to isn't a fill, but individual lines.


Much like how this uses a ton of pen dots. And again, even the portraits that don't hand-dither like LSL3's 435 end up looking weird because views as face parts.

[Kawa]

I have a persistent issue with pasting graphics that has finally gone too far to ignore.

This is the new closeup for Alhor[/url] in The Dating Pool. I imported this, then extracted the actual palette for use in Paintshop. I applied this exact same palette to the separately-cut out mouth and eye frames, and copy-pasted them back into SCI Companion.

Look closely. See how the white part around his mouth is redder than the rest? How there's bits of global palette in his nose? The same thing happens with the eyes, turning them from olive to brown and making the blink look particularly as if there's only 32 or so colors.

Even though I guarantee you every single RGB value in that source pic I'm pasting is in the actual, current palette.

I have a worse example, (palette included for illustrative purposes towards Letrune) but I can't bear to link to the version with the sprite on. Suffice to say it comes out looking so bad.

[troflip]

You didn't describe your exact workflow, so it's hard to know what the problem might be.

What options did you use for mapping the palette? Where are your images from? Paintshop? SCI Companion? In-game?

What does SCI Companion tell you about the palette indices that ended up being used for the view?

What happens if instead of pasting the image into Companion, you save it to a indexed-palette file (e.g. gif) and import it that way? A lot is lost when pasting images through the clipboard (transparency for one, and probably the 8-bit-ness too).

Can you post a comparison of how it looks in Paintshop vs how it looks in Companion vs how it looks when you paste it into another app (e.g. mspaint)

I tried importing your image into a pic, then copying the resulting palette into a view.
Then I pasted the image into photoshop and cropped it down to the mouth and nose. Then I pasted that into the view, choosing to map it to the custom part of the palette. Looks identical to me.