I find myself in the position of trying to consolidate some code, ending up with a function declared as follows:
int process_spell_target(int who, int what, int y0, int x0, int y1, int x1, int spell, int level, u32b flg,I'm not sure whether this is morally the right thing to do.
int region_id, int delay, int damage_div, bool one_grid, bool forreal, bool player,
void retarget(int *ty, int *tx, u32b *flg, int method, int level, bool full, bool *one_grid), bool *cancel)
This is a pre-release of 0.6.3. It should be stable, and just requires extensive playing to sanity check all the changes and find weird edge cases. Unfortunately, I had to cut trap regions before release, as this would otherwise extend an already late release unnecessarily.
Please see the official Unangband site for details.
I find myself agreeing with Eskil Steenburg's advice on how to learn to program:
I almost never use linked lists (I think there is one place in all of LOVE), and instead I try to use arrays, they are faster, smaller and more cash coherent.[Edit: Reading the article, I believe Eskil is primarily referring to linked lists where each member of the list is dynamically allocated and pointer referenced. It's possible to do an array based implementation of linked lists which avoids the issues highlighted in the comments, and then to infrequently expand the array using realloc in the event the array fills up. Angband uses the same approach, but usually has a hard limit on the size of each array, instead of dynamically expanding it].
This is the 500th blog post on Ascii Dreams - an achievement I've been waiting to reach for a little while now. I hope you'll excuse a little self-indulgence at this milestone.
I started Ascii Dreams following the advice of a number of people on rec.games.roguelike.development that I should post my 'design diary' notes in a blog, instead of spamming the newsgroup with posts. My first article generated what I feel was a positive, albeit long thread, spamming the news group with posts discussing classes versus skills, with my position opposed to both. I would also like to think it indirectly inspired Ironband, an Angband variant by Antoine - the first person to reply in the thread:
There are no classes in Ironband - all characters can fight, shoot, sneak and cast spells. You can specialise by raising the appropriate stats (e.g. Intelligence and Wisdom for casters). The other stats are Strength, Dexterity, Constitution, Agility, Stealth, Perception and Luck.It seems inevitable at some point as a blogger, you begin to worry more about traffic stastics than the quality of your writing. I achieved some early infamy through blogging, with articles being linked to by notable games journalists, up and coming games journalists and getting Slashdotted. But the Ascii Dreams traffic has remained consistent at around 150-400 visitors per day as opposed to experiencing the exponential growth which would support a web based advertising model - much to the disappointment of my wife. Unsurprisingly, the top read article is The Death of the Level Designer - a title I only partly regret choosing. More surprising, or perhaps not depending on your understanding of the human condition, is the popularity of the article on how to play different region DVDs on your Macbook:
| Page | Pageviews | Unique Pageviews | Avg. Time on Page | Bounce Rate | % Exit | ||
|---|---|---|---|---|---|---|---|
| 1. | 57,894 | 47,101 | 00:03:43 | 74.40% | 68.10% | ||
| 2. | 27,803 | 25,866 | 00:04:44 | 85.67% | 80.53% | ||
| 3. | 9,916 | 8,090 | 00:04:00 | 62.68% | 54.21% | ||
| 4. | 8,950 | 8,366 | 00:03:39 | 89.93% | 88.73% | ||
| 5. | 5,870 | 4,697 | 00:03:00 | 42.85% | 34.12% | ||
| 6. | 4,987 | 4,621 | 00:02:48 | 86.17% | 84.16% | ||
| 7. | 3,763 | 3,494 | 00:01:33 | 80.00% | 78.58% | ||
| 8. | 3,481 | 3,041 | 00:02:50 | 51.39% | 40.94% | ||
| 9. | 3,348 | 2,577 | 00:03:34 | 80.91% | 73.45% | ||
| 10. | 3,325 | 3,086 | 00:04:37 | 75.16% | 49.98% |
For those of you obsessed with A* pathfinding, you'll appreciate the cold water poured on the idea this thread on rec.games.roguelike.development. And then take yourself straight to this collabourative diffusion example before you get obsessed by the trap of perfect pathfinding.
...of the Interceptor.
If you're also developing an X-Com roguelike in secret, it's time to confess and share your hidden desires with other roguelike developers on this rec.games.roguelike.developer thread.
Too many game developers fall into the trap of making warriors re-skinned wizards: that is, the warrior class ends up have a list of special moves that in most respects acts identically to a magic user's spell book. There may be some overt themeing: the warrior may end up using themselves as a weapon, with more of an emphasis on buffing, but in all other respects, the abilities fall into the same categories that I talked about in the Designing a Magic System series of articles - the chief culprit of which is a screen or menu showing a list of abilities to pick from.
There are several ways around this. I've talked previously about using positional game play to trigger warrior abilities - by having special moves activate when a warrior moves twice in a single direction (charging), moves laterally (dodging) or stands still (blocking). Flend's DDRogue, reviewed by John Harris in his latest @ Play column extends this concept to a host of different special moves which are triggered by a combination of single moves. The weakness of this approach, as John points out, is that it is very hard to convey where in a move the player is currently using the user interface of a roguelike. In a fighting game, 'subcell stance' - such as whether a fighter is standing, crouching or jumping, is intuitively conveyed using graphics. A roguelike designer does not have this luxury.
Another key issue is a warrior's limited ability to flee a fight. Mages are given an automatic out in the form of teleport spells, which allow them a variety of tactical options for moving around the battlefield. To duplicate this in Unangband, a warrior ends up carrying a large number of different items - as highlighted by Bandobras in a recent Unangband ladder dump - which drags them back into the pick items from a list camp of game play.
I've tried several approaches to this. My initial thoughts were around letting the player manipulate whether they wanted a half-cost move versus a full cost move. A half-cost move - to reflect running around the battlefield - would allow the warrior to move around dynamically but incur a movement debt which would have to be paid off later. This movement debt appears in Unangband currently as fatigue, which slowly accumulates if the player is moving unusually (through water, mud or other terrain; highly encumbered; hasted) and encourages the player to periodically rest by causing them to eventually faint if they do not do so.
The problem with half-cost moves is the user interface overhead in order to be able to take advantage of them. I initially went with a modal approach, where the player would always prefer half-cost moves unless searching, which made the mental book keeping required to move far too high - you'd have to rest far too frequently while simply moving around the dungeon if you were in the wrong mode. A bucky key approach, where a modifier key such as Ctrl or Shift is used to indicate a half-cost move, overloads the user input requirements.
I like the semi-realtime approach that Wayfarer uses, simply because it emphasises mashing the keys to move quickly. From this post on rec.games.roguelike.development:
Basically yes -- it "feels" real time because different speeds areUnfortunately, messing with timings in a roguelike is something that has to be done very carefully, as time sequence is something core to the balance of the game.
shown as different npc-slide-rates, and they complete their current
move if you idle. So you can outrun a glormbeast if you're holding
the movement key down, but if you're doing a tap-think tap-think
process it will stay at your heels.
Posts
All Comments
