God, this one was definitely the biggest headache of all of the
magic number erasing. Never did I expect such cryptic problems.
The first problem was that the entirety of the player's weapon
struct was a part of the save file, *including the weapon's "image
indexes"*. Since the indexes have been changed, and the originally
used one is now unavailable when it's requested, this was causing
a segfault later on. Had to fix this by setting the image index
when the game is loaded.
The second problem was related to another bug I've been confused
about for years: the one that causes mobile rays to fire 5 green
shots. The entire reason those shots were green was because
the weapon's image indexes were undefined, and *that was causing
them to default to 0*. 0 was simply the index of green plasma.
Of course, though, now attempting to use that image causes a
segfault, so for now, I've fixed this by changing the image index
of the mobile rays to the red plasma bolts.
There are still some magic numbers left, related to the intermission
screen. But the hardest part is now done, thank God.
I found myself confused by the interception chances, because it looked
like the chance of interceptions was decreasing slowly over time.
Turns out, it was the denominator in a fraction; a "chance" of 300
meant that there was a 1 / 300 chance.
These are just there so that you can define these at compile time
without modifying the source code. In the future, I would like this
to only be the *default* screen width and height, because I'll want
to add an option in the menu to change the resolution from 640x480
up to whatever your monitor's resolution is.
This is capable of handling any resolution which is at least 640x360.
Actually, resolutions a bit smaller than that can be handled. No
manual adjustment is necessary. :)
Now, the intermission screen needs to be fixed...
This places the logo at 1/3 of the way down the screen, rather than
half of the way down, and the menu ends up about in the middle of
the screen now. This doesn't really matter as it is, but it should
help some with lower resolutions.
Also, now that the magic numbers are gone, I was able to get rid of
all the hardsetting of numbers for the special ship indexes (which
was only in place to avoid breaking the magic numbers).
This name is better since the difficulty isn't actually *exactly*
like the original; it's just trying to be as similar to the classic
experience as possible.
Like "scripts", these definitions in text files make absolutely no
sense. They are completely unreadable, and Starfighter's engine is
inflexible anyway.
I'm going to do this for all of the stuff in the "data" directory.
It was obviously an attempt to make Starfighter more flexible
somehow, but it fails at that entirely. More importantly, these
things are both unreadable and easy to make mistakes on. Simple
C code is much easier to read.
The only disadvantage is that recompiling is now needed to change
the "scripts", but considering that they had hidden limits and
no one was making custom missions to begin with, I don't consider
this to be a real loss.
To my recollection, actually, this bug has *always* been present.
At first, I fixed it (the problem was the entry in script13.txt
was malformed), but when I did that, it caused the objective of
destroying Ursula's ship to be considered completed. In other words,
the game doesn't even have code to handle this as a failure
condition! So I've decided to remove it instead. It wasn't that
great of an idea, anyway.
I rewrote the method because what it was doing was so confusing, I
couldn't figure out whether it actually worked right or not. I think
it did, and it's only 2/3 of a milllisecond anyway (not noticeable
at all), but this new way of writing it is much clearer.
It wasn't very useful. All it did was offer the option to center
text and take away the option to wrap text. I've moved the text
centering option to gfx_renderString and replaced all uses of this
function with that function.
The fourth is simply a duplicate of ship_collision for bullets. A
bit redundant, but I figure it's clearer of a definition. Besides,
this opens up the door to possibly making bullets a different struct
type in the future, if that turns out to be desirable.
onpon4