Sturm Und Drang IF: A Stormy Romance

The Sins of Inform 7

October 14, 2010 at 2:23 am (Inform 7)

Inform 7, while powerful, has some serious flaws that a round or two of proper beta-testing from anyone outside the I7 sanitarium would have discovered. But as Inform has always been an incestuous affair governed by the Old Ones, it’s no surprise that horrific wonders would be birthed. Behold, the sins of Inform 7.

1. Tables. In Inform 7, tables are a powerful construct, allowing more than one interesting use (such as responses given a proper topic). However, when tables fall apart, it’s not pretty and it’s downright counter-intuitive and painful. Where would that be? Where the comparison between a table in a book is at its weakest, because a table is wrenched into a programming context, but retains all the restrictions of being a table in a book. Yes. You too can try to fit paragraphs of text into narrow columns!

In other words, this makes sense:

Table of Outer Planets
planet   	semimajor axis
Jupiter   	5 AU
Saturn   	10 AU
Uranus   	19 AU
Neptune   30 AU
Pluto   	40 AU

THIS does not:

Table of Instruction Options
title subtable description toggle
“About Interactive Fiction” a table-name “The game you are playing is a work of Interactive Fiction. In interactive fiction you play the main character of a story. You type commands which determine the actions of the character and the flow of the plot. Some IF games include graphics, but most do not: the imagery is provided courtesy of your imagination. On the other hand, there’s a wide range of action available: whereas in other games you may be restricted to shooting, movement, or searching items you can click on with a mouse, IF allows you a wide range of verbs.” a rule

It’s ugly, hard to follow, and loses all of the power of the metaphor. It really forces people into defining the table rows later where text can flow freely, which says “chuck it” to the whole table layout.

2. Documentation. I don’t often toot my own horn, but I write documentation FOR A LIVING. I get paid to understand how to fit docs to audience and I understand genetically the need for testing. If the Inform 7 cabal could have been bothered to ask for public feedback (something like “Hey, we’d like input on the docs. Can anyone help?”), I could have spared them tons of harsh feedback, and a general decline in their estimation in the whole IF community.

The documentation for Inform 7, is in two words, an obtuse horror. It begins with a trivial example, but never builds on the trivial example to a robust one. Instead it leapfrogs from application to application of the general concept. It’s much like starting off in a field behind your house, and within ten minutes of walking, you’re on the moon. Newbies it doesn’t help; old hands it frustrates. The docs are simply masterful at giving no-one what they really want.

For instance, say that you want to set the describe properties of a thing. Specifically, you want what displays when the object is in a room to vary. Good luck on finding that. The docs never mention it. Sometimes the only reference to how to do something is buried within the documentation of particular extensions. Yeah, I kid you not.

Mostly, the docs describe how to do little things that don’t add up to much; even the extended examples aren’t thorough. I’m sure I’ll find out just how threadbare they are as I go searching through the entire freaking book just trying to find a clue. It’s documentation by piece, but not by whole; I don’t think anyone reviewed the finished project as a whole.

The docs read like some barely-connected daydream instead of an actual description of how to do things or even, how things work. It’s almost like the authors weren’t well-versed on the necessities of typical IF games and the needs of their authors; that sounds bizarre, but how else can you explain the patchwork effect here? I suppose substance abuse is always a possibility, as is the authors being so far out-of-touch with other human beings, that they really believed that everybody knows everything they do and so “Inform for Me” was all that was necessary. What we really need is an “Inform 7 Beginner’s Guide” to parallel the magnificent “Inform Beginner’s Guide” put out by Roger Firth and Sonya a while back. That book’s approach covered all the bases and had a handy set of appendices that listed common commands, properties, and so on.

3. The language itself. I7 is in many ways, a real step backwards for people who actually want to make games, instead of people who want to throw together a few cute objects that don’t do much. Take for instance the age-old problem of objects that are both containers and supporters, like oh, say, a toilet bowl. Is I7 any improvement over I6? No. You’re still stuck trapping verbs that go to one object and shunting them over to another. You’re still stuck modeling two different objects and trying to obscure their differences to the player.

The fundamental problem is that the language just isn’t thing-oriented. It doesn’t help you make things and govern their interactions well. It’s too high-level to deal with the messy reality of a world of objects, and it makes the low-level stuff even more difficult by forcing you to write near-English sentences about stuff instead of

breaker.oos = true;

The latter is more compact and really, semantically more sensible, as you don’t talk your programs into existence; no, you program them, and a programming languages are created for the purposes of achieving what is cumbersome in English. C, PERL, JAVA, all use these “linguistic shortcuts” or more properly said, have their own language constructs to achieve things simply and quickly. I shudder to think of trying to do pointer arithmetic in I7.

But really, the source code of Weishaupt Scholars says it all. I remember beginning to program in I6. I checked out the source code to Christminster. If you were trying to learn a new language and you came across these two examples, which do you suppose would be easier to learn?

No matter how much I7 repents, it can’t be saved.