Month: September 2020

Posted on

ACotGK: Quest Order Bug

Found a “neat” bug today, replaying Chapter 1 for testing purposes. Here’s what happens. The party enters the final fortress for Chapter 1, and is presented with a locked door in front of them:

Fortress entrance.

On clicking on the locked door, “You need the Bagduul key to unlock this door.” Not remembering the exact details, I figured I had missed the key somewhere, so I went through the previous few areas looking for the key. Nowhere to be found!

So, I went back to the source code. You are supposed to get the Bagduul key from the big boss of Chapter 1. Then, you can open the treasure room (using the key), where you get your well deserved reward. But, in this case, the treasure room was spawned as the very first room in the fortress, so you couldn’t get to the boss to get the key in the first place. Oops!

My first fix was to make sure the treasure room wouldn’t get spawned as the first room in the dungeon. But then I had a thought, borne out by some testing. The treasure room could still get spawned as the room right _before_ the boss room, again blocking access to the key need to access the treasure room. Oops again!

At least I thought of this before it caused more problems. The fix I settled on was to move the treasure room to an attic, a mini-level upstairs from the main level. The locked door is then on a completely separate level, and can’t possibly block access to the boss.

This does show one of the drawbacks to procedurally generated levels. Just because the level generation code makes a good level once, doesn’t mean it always does. As a coder, I need to make sure to test things in as thorough a manner as possible, and also keep the levels (and quest logic) relatively simple so there aren’t so many ways to screw it up.

Posted on

The Infocom Project: The Witness

Cover Art

I haven’t done this in a while. In fact, I’ve switched blog platforms since the last one I did! You can read about my project to play all of the Infocom games back here on GameDev: http://www.gamedev.net/blog/717/entry-2259971-the-infocom-project/

Up this time, “The Witness”, one of the earlier Infocom games, the second in their detective mystery games (following Deadline).

The good:

  • The writing and atmosphere. Detective noir fiction, but in adventure game form!
  • The mystery. No spoilers here, but I thought the overall mystery made sense, had twists and turns (but not ridiculous), logical, and ultimately fairly satisfactory in conclusion.
  • Contained location. Usually I make a map for Infocom games, but this time I just never got around to it, because I didn’t need to. The location, small for an Infocom game, was straightforward, and I quickly internalized where everything was.

The bad:

  • Mystery as text adventure. I disagree with many of the reviews I’ve seen; I don’t think the “whodunnit” mystery worked well in the text adventure format. In particular, the interaction of the “figure something out” mechanic with the “save/restore” mechanic. I’ll give an example. Part of successfully completing the adventure is to determine the motive of the murderer. I managed to do that, but on completion of the game, I was informed that I hadn’t adequately figured out the motive. I restored and went back through, trying all sorts of new things with no success. Eventually, I went to Invisiclues, and realized I had solved the motive properly, I just hadn’t done it in the particular save game that I took to completion. I found this obtuse and frustrating.
  • Human characters. Infocom games don’t deal well with human characters; people are just too complex and creative to fit well into the strictures of a text adventure. Communicating with them becomes an exercise in frustration as you try to determine just the right way to ask them about something they might not even be programmed to know anything about. You can argue whether this is due to the technical limitations of the early text adventures (128K of memory, I believe), or whether it is just too hard to program realistic humans. But in either case, I find that humans in Infocom games only work well when they have a very limited and obvious role. In this case, you have full conversational access to a number of human actors, and it doesn’t work for me.

Puzzle frustration: Low. I was generally able to figure out the mystery and what was going on. My frustrations were more with getting the game to recognize what I already knew. I did refer to Invisiclues several times, but I generally came away with, “I already knew that!”

This is my least favorite Infocom game thus far; t just didn’t work for me. Other reviewers have different opinions, so this might just the limitation of me, playing this game 40 years out. Jimmy Maher review the game here: https://www.filfre.net/2013/03/the-witness/

I’m not sure what I’ll play next, but I’m going to try to get to it before years have passed.

Played Games, Best to Worst:

  1. Enchanter
  2. The Hitchhiker’s Guide to the Galaxy
  3. Sorceror
  4. Zork III
  5. Zork II
  6. Zork I
  7. Planetfall
  8. Trinity
  9. Wishbringer
  10. Stationfall
  11. Spellbreaker
  12. Plundered Hearts
  13. The Leather Goddesses of Phobos
  14. Infidel
  15. Starcross
  16. Seastalker
  17. The Witness