Please help with testing the upcoming ScummVM 2.6.0 release!

General chat related to ScummVM, adventure gaming, and so on.

Moderator: ScummVM Team

Roody
Posts: 4
Joined: Sun Nov 27, 2016 2:45 am
Contact:

Re: Please help with testing the upcoming ScummVM 2.6.0 release!

Post by Roody »

GusCE6, for your purposes, I think you would be better off using the Agility interpreter. I don't think it claims to have 100% support for all AGT games (since certain aspects of old versions of AGT have never been fully documented), but it is the best experience for playing AGT games outside of a DOS window.

Gargoyle, by design, intentionally ignores things like color choices. It's basically like an e-book reader for interactive fiction, prioritizing a comfortable reading experience over fully emulating these systems.
User avatar
TrivialBalderdash
Posts: 10
Joined: Tue Aug 31, 2021 3:45 pm

Re: Please help with testing the upcoming ScummVM 2.6.0 release!

Post by TrivialBalderdash »

I completed a play through of the CD version of Tender Loving Care. I was running the June 13 stable build of 2.6.0 on MacOS 11.6.7 on x64 Intel.

I ran into a few issues -
  • Several crashes during the surveys, in Chapters 11 and 15 - 13624
  • During a few movies, the audio and video were out of sync by up to a second
I never played the game when it was originally released. I actually hadn't even heard of it until it was in development for ScummVM. I picked up a copy on eBay to complete the Trilobyte collection. So I don't know if the A/V sync was an issue with the original. I do have a 486 and Pentium 1 I could go test on if needed.

I was able to see two of the 6 or 7 endings. Anyone know how to test for the other endings - is it random or based on the survey responses?
User avatar
TrivialBalderdash
Posts: 10
Joined: Tue Aug 31, 2021 3:45 pm

Re: Please help with testing the upcoming ScummVM 2.6.0 release!

Post by TrivialBalderdash »

I completed a play through of Clandestiny. I was running the June 13 stable build of 2.6.0 on MacOS 11.6.7 on x64 Intel.

I saw two minor issues -
  • Screen offset with in-game save/load screen - 13636
  • There was choppy playback on some of the movies. There would be a "blip" in playback every 5-10 seconds, but it wasn't on every video and it wasn't consistent. More testing is probably needed. I wasn't able to replicate this reliably, so I didn't put in a bug report.
I played the game on the "Brave" difficulty setting.

Overall, it was a great experience replaying this game, and so much better than on original hardware. I had forgotten just how much CD load time can make the whole game feel sluggish.

Interesting side note, four of these puzzles were re-released as part of Uncle Henry's Playhouse/Mindblower. They were updated slightly with different sound effects. As I was playing Clandestiny, I noticed missing sound effects for those puzzles and assumed it was a bug. Turns out, those sounds weren't in the original game.
User avatar
szczuru
Posts: 17
Joined: Sat Aug 05, 2017 3:17 pm
Location: Poland

Re: Please help with testing the upcoming ScummVM 2.6.0 release!

Post by szczuru »

Is there any way to enable logging in PSVita version? I've tried to run Sanitarium which is working nicely on Winx64 but on Vita i've got black screen (without freeze 'cause i still can access to scummvm's pause menu or debug console).

Both (win and vita) versions are df2121b0.
Also tested Spider-Man Sinister Six which seems to works well on PC and Vita.
GusCE6
Posts: 62
Joined: Wed Nov 07, 2018 9:50 pm

Re: Please help with testing the upcoming ScummVM 2.6.0 release!

Post by GusCE6 »

I've done some more testing with the Gargoyle Interpreter (2022.1 Windows i686). Got to contribute SOMETHING here besides links to SwiftShader even if SwiftShader is excelent...if you can't get SCUMMVM running no matter what you've tried that might be the answer. Just please be certain any future versions can use DirectX as well as OpenGL. ^-^
This coincides with more learning efforts with AGT.
I've also been using DOSBox 0.74 for a game I'm programming. Hence DOSBox, Gargoyle, and SCUMMVM are being all tested.
In a nutshell if a future version of SCUMMVM wants to handle AGT properly it should combine the best aspects of both Gargoyle and DOSBox because neither one gets it 100% right, but each can do something the other can't.

Earlier I mentioned how keyboard images do not come out right with Gargoyle because it is not wide enough, but unlike DOSBox one can just manually adjust the window size or better yet use the Gargoyle INI file to make 80 columns the default- problem solved. SCUMMVM seems to allow manual adjustments at this time.

DOSBox allows AGT to use sounds- Gargoyle and SCUMMVM do not allow the sounds to work.

DOSBox uses the colors set in the TTL file and they can be changed manually by the player during the game. With Gargoyle the colors cannot be changed manually in-game but many more color options are available via the INI file. It is even possible to have keyboard graphics and keyboard animated sequences a completely different color although it must always remain that color.
SCUMMVM is always black text/white screen. I put the altered INI file in the SCUMMVM folder as well as the game folder- no effect. DOSBox and Gargoyle are superior in that way.

Gargoyle has some serious problems- for example, in one game I am programming to escape from the first room some cushions must be searched to find a key. Here is the meta-command:

COMMAND SEARCH CUSHIONS
AtLocation 2 (* Starting ROOM 2. This is a very specific command! *)
FlagOFF 4 (* CUSHIONS NOT SEARCHED YET! *)
PutInCurrentRoom 321 (* BRONZE KEY IS PUT IN ROOM 2 *)
PrintMessage 56 (* BRONZE KEY FOUND! *)
TurnFlagON 4 (* KEY found, this command will be skipped next time! *)
DoneWithTurn
END_COMMAND

What you get is "Don't know how to search here..." so it is literally impossible to get out of the first room. Note that the cushions are actually in Room 2. Also note that SEARCH is set as a "dummy verb" so it is valid.

I use generic replies to certain inputs- for example, EXAMINE WALLS always yields one cover-the-bases reply but what you get is "You don't see any walls here." My guess is that since the walls are in "Room 0," a sort of Limbo with AGT, the noun exists but is not present- so Gargoyle cannot handle that. The simple meta-command is this:

COMMAND ANY
NOUNIsNumber 308 (* WALL *)
PrintMessage 30 (* GENERIC WALL STATEMENT *)
DoneWithTurn
END_COMMAND

"NOUNIsNumber" means just that- if the noun in the input is 308 (WALLS) then print that message. Gargoyle can't do it but it can handle nouns which are actually in the room (e.g. the cushions). DOSBox has no problem here.
The same problem exists if WALLS are the object- "UNLOCK DOOR WITH WALLS" won't work the way it does in DOSBox (yes, I have a mocking message for such commands...).

QUICK NOTE: AGT handles commands as "Verb-Noun-Preposition-Object." Adjectives can be included.

Programmed meta-commands in those games do not allow a character to do certain things like throw amulet or drop another certain item, but with Gargoyle or SCUMMVM you can- and in some cases it is not possible to pick it up again.

JUMP yields "Don't know how to jump here..."

COMMAND JUMP
PrintMessage 54
DoneWithTurn
END_COMMAND


Input "yell" and normally one of two things can happen at random: one, you are simply chided but the other is that you are heard and captured game over. This is what you get instead:

> yell
GAME ERROR: Invalid argument to metacommand token.
YAAAAEEEEEEEEOOOOOOUUUUUAAAAHHHHHH!!!!!
GAME ERROR: Invalid argument to metacommand token.

The "Yaaaeeooo..." isn't all of it either.

Gargoyle has a number of problems when meta-commands are involved. The GAGS game "Alice" ran fine since GAGS does not have meta-commands. The weird part is certain meta-commands DO work; trying to kill a monster when you do not have the needed item(s)* or do not specify what to use will yield one of two random messages- yet a simpler effort fails. If you use the proper or wrong item either way everything works.

Testing with SCUMMVM 2.6 shows it behaves like Gargoyle but without the extra INI options.
AGT Master Edition has illustrations (PCX), but these cannot be viewed with Gargoyle or SCUMMVM. These are not problems with DOSBox.

Indentations do not work with Gargoyle or SCUMMVM. Alterations I made with the RUN.EXE file have no effect (e.g. "Instructions" should be "Introduction"). That made me suspect that AGT's RUN.EXE is not used; removing it proves this. In the title screen text which should be centered is not. Could someone link me to Gargoyle's "AGT-RUN" source code for comparison with AGT's own?

The only (albeit serious) problem with AGT using DOSBox is the "AskQuestion" meta-command. It works, except simply hitting ENTER counts as answering the question properly every time- there is no way around that. Also, after the first time trying to answer a question hitting SPACE causes the cursor to skip a space. Gargoyle and SCUMMVM do not have these problems.
I'm guessing it's the meta-command "AskQuestion" since that skip-space problem happens before you even try to ENTER the answer.
Note that both of these particular problems exist with Puppy Linux DOSBox and the two mobile versions for Windows CE PocketDOS 1.2.1 and PckDOS 0.63.
I
THINK
IT
IS
ORANGE

instead of

I THINK IT IS ORANGE

A problem with Gargoyle and SCUMMVM is if you enter "I think the answer is orange" you will get the proper reply but this as well: "[i -> inventory]"

Entering "HELP" does show the extra tip or description but it also shows this: "[Try ’glk help’ for help on special interpreter commands]."

I have downloaded the AGT source code and with luck can maybe fix the problem at the source after learning the needed PASCAL. Due to how DOS emulators are just about everywhere- including a puny Windows CE netbook- fixing that problem so it works on DOSBox, even if it messes up the game for Gargoyle, is preferable.
There it is. If SCUMMVM can run AGT like DOSBox but handle question/answer like Gargoyle it will be better than either one. Since the Windows version of SCUMMVM works with Puppy Linux via WINE 1.7+ that would work too. Of course if there is an actual AGT interpreter and it runs via DOS so DOSBox can run it indirectly then maybe that will do.

Note that I am using AGT "Big," which is AGT 1.7 except it allows many more rooms, nouns, commands, messages, and variables. This is the version I want to fix.

By the way, when you say "source code" for the game "Foxtail" do you mean Unity or for the game itself? If the latter I'll ask the programmers over at Steam. Where would they send it? If the game will always run on SCUMMVM and there is a version of SCUMMVM for just about anything then they can make more money and that would be a big help for them given what's happening over there.


* With GAGS and AGT upon encountering a monster you cannot leave the area and only one item can kill it. Using meta-commands I have it so you can leave the place and more than one item can kill it. Heck, you could have hit points and even special abilities. If I can figure out the source code maybe I'll also try to include color changes among the meta-commands so keyboard images can be more like "SanctuaryRPG Black;" e.g. if the text is normally green and you encounter a red fire dragon the keyboard image and text would be red- afterwards back to normal. If you go through the cave into a desert the text can be yellow, or maybe yellow screen/black text. A watery cavern, blue. An illustration for sky, at night black screen/ white for stars or daytime blue screen/white for clouds. You get the idea. Maybe add colors to the palette?


I'm having trouble using the Turbo PASCAL Compiler and the DOS version of Agility.
GusCE6
Posts: 62
Joined: Wed Nov 07, 2018 9:50 pm

Re: Please help with testing the upcoming ScummVM 2.6.0 release!

Post by GusCE6 »

GARGOYLE TEST UPDATE 2

First, please keep in mind that, due to situation and finances, I do not go out often and must use public wifi. My last post was over a week ago Friday- the last time before that was maybe Monday a week and a half before. As a result I have to ask things while online before being certain I cannot handle it myself because the possible time lag makes normal ask-answer impractical. I am sorry about any frustrations that may cause here.

You know I've been testing AGT with Gargoyle, SCUMMVM, and DOSBox; as of that Friday I also downloaded AGiliTy, several versions, including the DOS version. Hopefully this can help here.

Overall DOSBox has proven superior to the others, except for the AskQuestion problem. Well, Friday I also downloaded Turbo Pascal and the source code for AGT. The instructions were somewhat lacking so it took a bit of time, but the basics were finally figured out. Now source code could be compiled into a RUN.EXE file, happily the Big version no less.

I do not know PASCAL but it does somewhat resemble Javascript and in some ways Commodore-64 BASIC (yes it's been around that long...get off my lawn). Using Notepad++ I found the code for AskQuestion and, after considerable tinkering, managed to solve the Just-Hit-ENTER-And-It-Counts-As-The-Right-Answer problem. The problem of anything after the first time going downward

HE
IS
A
SHELTIE

instead of

HE IS A SHELTIE

is still there but THE big problem is solved, and as long as the answer is just one word there is no problem at all. It seems as though the go-down problem does not affect the validity of the answer, but the word must be alone- SHELTIE is fine but SHELTIE! is not.

(NOTE: If after typing in the first character you hit BACKSPACE then type in the sentence it is on the same line. A clue.)

Curiously what I did somehow causes AskQuestion to malfunction on Gargoyle, AGiliTy, and SCUMMVM. However, since there is a DOSBox for Puppy Linux, Windows, and even Windows CE- and WINE alows use of the Windows version on Linux- the bases are pretty well covered. Those others also malfunctioned with other aspects of my programming before any alterations.

Along the way I also converted DELAY into milliseconds instead of seconds. Can't do anything with colors yet, including the curious inability to make the background yellow. Maybe the sixteen colors given is the limit?

I've tried several games with Gargoyle and AGiliTy and for the most part they work, Gargoyle's INI file a great feature, so it's my programming style that must be causing the problem. If anyone here would like the test game to study for SCUMMVM and AGiliTy let me know where to send it.

It was satisfying actually solving such a problem myself, even if low-level.
GusCE6
Posts: 62
Joined: Wed Nov 07, 2018 9:50 pm

Re: Please help with testing the upcoming ScummVM 2.6.0 release!

Post by GusCE6 »

There is no question that neither Gargoyle nor AGiliTy can handle the way I program AGT games, only DOSBox can, even though what I do is legal. Does anyone here want me to e-mail or link to the prototype game so far for analysis?
Post Reply