I'm annoyed by everyone calling the wiki outdated!

[tasonir]

We're updating it every day! No really, here's the changelog: http://crawl.chaosforge.org/index.php?t ... entChanges

Averages around 10 or so updates a day lately; the goal of this post is to triple that! I've been checking all of the pages linked from the main page, and they should all be updated for 0.10. Some questions:

Is it still true that spellcasting takes 130% exp to raise, and stealth/invocations/evocations takes 80%? I believe that has been removed and all skills are the same. It is mentioned here for updating once I am sure of the answer.

I added a link to the knowledge bots on the Monster page - there was talk about getting a script to update all the monster stats on the wiki, but I don't know how to run it, so until then, linking to the knowledge bots is fine, imho. Honestly every monster I've ever looked up on the wiki has correct stats aside from some of the statues, and I could probably add those by hand.

I'm tempted to recreate the most awesome flowchart ever and just change "MDFi" to "MiFi" (flowchart found at http://crawl.chaosforge.org/images/d/d2 ... wchart.jpg)

Other than that, the only main pages that really need work are some of the race pages. Draconians in particular have been changed around a bit recently and I'm just not experienced enough with them to update it as a whole, although I did make some corrections. Merfolk is flagged as version .8, but it seems pretty good to me. I haven't checked all the aptitudes, though. Spriggan is likewise flagged as .8, and mummies as .9. Both seem pretty well written, but post here if you see anything inaccurate on those pages, and I'll change them. You are free to rewrite it yourself if you have an account, of course.

Honestly, it's looking fairly accurate to me. Today I only checked over the "main" pages so if there's inaccuracies in some deeper pages please feel free to mention them here as well.

Yes, i did leave in the line "A Spriggan with a rod of destruction is nigh unstoppable, at least by monsters." This has been quoted by cerebovssquire as being laughable, but he didn't quote the rest of it: "The problem is to survive until you can acquire or find (good luck with that!) one, and then to survive the hunger from the rod (Int and spellcasting skill help with spellcasting hunger, slow metabolism 3 helps with walking about, but NOTHING in the game completely mitigates Evocations hunger.)."

I think the complete passage is accurate information. If it really bothers you it can be removed, I suppose.

[rchandra]

experimentally, stealth still goes up faster than, say dodging (Spriggan to compare), which matches the changelog ("no longer heavily discounted" suggests it's still discounted). I seem to remember it used to take 50%, so 80% is probably right.

skill pages like Invocations refer to an outdated version, but I think their data is still correct - except that MD is still there, and Op/Te have errors. Not sure what the source is, looks like it might be autogenerated from something.

Thanks for updating it, wikis may never be perfect but they can easily be pretty good.

[tasonir]

updated the invocations page at least then, I should get around to checking other skill pages soon. I left MD in, the wiki generally likes supporting historical information, even though I'm not personally too fond of it.

[galehar]

There. Spellcasting is still 130%, and evoc and invoc are still 80% and stealth has been raised to 80% too.

There is a script to keep the knowledge bots monster stats up-to-date. Maybe it could be adapted for the wiki. |amethyst and kilobyte are maintaining, so if you're interested, you can ask them for help on ##crawl-dev.

You don't actually need to wield a weapon to train its skill. Carrying it is enough provided you can wield it easily (which means what your wielding and the weapon carried must be known uncursed and not vampirism/distortion).

Thanks for doing this!

[XuaXua]

4 times out of 5 someone is sitting on them who will revert every change to "their" guide, so even if we put in the effort to rewrite them it'd all be wasted.

Wiki history. Revert the revert or at least spellcheck.

[Blade]

And then they'll revert the reversion of the revert. Why bother fighting with idiots when they're so intent on defending their idiocy?

Minmay is completely correct. There is far too much subjective information on the wiki, and it can easily lead new players into following bad advice religiously.

The other problem is that Crawl has a fairly small playerbase and gets updated every day. There is so much to keep up with that it's unlikely that everything will ever be up-to-date (and, given the history, unlikely that people would be able to trust it even if it was).

It's not a lost cause, but there will always be problems with it. Updating it is worthwhile but does not make everything magically perfect.

[mageykun]

Maybe I'm naive, but I would think that the whole reversion-loop could be cut short by some sort of administrator or moderator power. Doesn't wiki structure allow that? But I'd second the idea of compartmentalizing subjective strategies and opinions vs facts. Put the guides in their own place, away from factual articles, and put a big disclaimer above them.

I'd say the biggest potential strengths of the wiki over the knowledge bots is format and size. For instance, there's space in a wiki for detailed, complete comparison tables. This for example, I consider one of the more useful references- it allows quick comparison of weapon types and allows you to make easy, informed decisions on base type based on damage, skill investment, and handedness. There's also room to explain things in detail that the bots gloss over quickly, or assume you understand. God pages are a good example of this- wiki pages have the potential to be very complete, while bot entries are brief and segmented into separate entries for the <god>, <god> reasons, <god> wrath and sometimes special abilities. The wiki list of all mutations is another potential example of a more complete list with detail, although that's in need of improvement.

So while the bots make a very effective stat-book (possibly an annotated stat-book), and will always be more up to date, what the wiki should shoot for is to be a report. The stats expanded, explained, and compared.

...besides that, I'd have to say one of my biggest frustrations with the wiki (and to some extent the bots) is the search functions. You almost have to know what you're looking up before you can look it up. It would be really handy if the search function offered auto-completes to existing articles, like the nethack wiki.

[tasonir]

We tell new players to not take advice from these sources because it is often bad. A lot of us actually do change factually incorrect wiki pages - but we can't touch the character guides and such because not only are they terrible, 4 times out of 5 someone is sitting on them who will revert every change to "their" guide, so even if we put in the effort to rewrite them it'd all be wasted.

Being outdated is a problem but it's not nearly as big a problem as this. I still like evilmike's idea to split off all the advice into a separate namespace.

I'd be fine with a seperate namespace for guides, I think they tend to mention tradeoffs/variants, or at least I tried to in my naga guide. It would be very easy to add a {{advice}} template which could be at the top of an article: "This article contains advice from other players, and may be subjective. Alter to fit your playstyle as necessary." I'm glad to have other people edit it, it still needs some polishing. The main thing I want to do is get much more links to other wiki pages, I didn't have the energy to go through and link to relevant topics when I posted it. That, and I'd like to put in slightly more heavy steering - I really think statue form is the best of the three listed transmutations, and should emphasize it, but I'll still mention the other options. Also I'm considering renaming it the naga melee guide, since I tend to be playing them as fighters these days

So since it takes 2 minutes I went ahead and created http://crawl.chaosforge.org/index.php?t ... ate:Advice Up to you guys how much we want to plaster it all over the wiki, but certainly character guides and some of the escaping from trouble, additional tricks, etc pages could have it.

[mattlistener]

Can anyone actually point to reversion-battles going on in character guides on the crawl wiki? I've updated a few guides based on my own experiences and information gleaned from developer discussions, and they've stuck.

[sanka]

Yes, i did leave in the line "A Spriggan with a rod of destruction is nigh unstoppable, at least by monsters." This has been quoted by cerebovssquire as being laughable, but he didn't quote the rest of it: "The problem is to survive until you can acquire or find (good luck with that!) one, and then to survive the hunger from the rod (Int and spellcasting skill help with spellcasting hunger, slow metabolism 3 helps with walking about, but NOTHING in the game completely mitigates Evocations hunger.)."

I think that this is still not very good advice, on my own spriggan experience. I'm a very bad player, but I'd like to gather some points, and you may see if I'm right that more experienced players will give other advice than you.

Hunger is rarely an issue. I very rarely has food problems with my spriggans, wether they use rods or spells, or not. (I did not tried a spriggan of Elly tough).

And +3 aptitude is nice, but since rods never fail, and for many spells the power is not so important, it's not really important to have a very high skill. If you cast spells, then you usually do not use too much rods, because spells are better. If you don't, then even a troll will have more than enough xp to improve evocations. Of course high evocations has an effect, but it's no way necessary to have very high.

The "nigh unstoppable" is not true because rods have very limited charges. They can be enchanted, and a spriggan can almost always run away to regen, but they are not really usable as the primaly damage source.

A spriggan with Orb of destruction however is really unstoppable. Conjurations is much more useful than evocation (if you do not follow Trog or Nemelex). Hunger is not really important.

[ebarrett]

If a guide is longer than the DDEE/DDNe guides in the learndb, it is either too subjective or bad. Another decent guide is tartakower MDFi, but only up to tartakower[1].

[Blade]

If a guide is longer than the DDEE/DDNe guides in the learndb, it is either too subjective or bad. Another decent guide is tartakower MDFi, but only up to tartakower[1].

(and ??hyperbolic isn't a half-bad spen guide either)

[galehar]

But I'd second the idea of compartmentalizing subjective strategies and opinions vs facts. Put the guides in their own place, away from factual articles, and put a big disclaimer above them.

Seems to me that strategic advices are already in a separate section on each page. And guides are also clearly separated on the main page. I don't think a separate namespace (as suggested by others) would help at all.
The "difficulty of play" section for species is problematic. It's in the middle of the page, surrounded by objective data, but it's a subjective and often controversial section. Not sure what to do about it.
A disclaimer could be good, but I'd put it right on the front page rather than spamming every single page with it. Something like:
"Strategic advices and guides found here are not always written by expert players. They may very well be dubious, outdated or plain wrong."

[flun]

Maybe I'm naive, but I would think that the whole reversion-loop could be cut short by some sort of administrator or moderator power. Doesn't wiki structure allow that?

So far there have not been any serious reversion battles. In the event that one comes up, there are two active users with admin access (MoogleDan and me) who can take care of it.

If anyone wants to help update the wiki but doesn't know where to start, there are a few projects listed here:
http://crawl.chaosforge.org/index.php?t ... t_projects

[XuaXua]

To avoid reversion battles, you can make gradual, one-two sentence changes, focusing on spelling or grammar and swap out the concepts you don't like in the meanwhile. Save and repeat.

[Bim]

I love the wiki. I've edited/added a bit to it (I FEEL LIKE INDIANA JONES WHEN I DO) and I think we should all be grateful - even if some of the stuff is outdated. To be honest, crawl evolves so fast that it would be impossible to keep up completely up to date, but I do agree that some more disclaimers added to the strat. guides/subjective material might be good, as some of them are a bit too dated/unclear.

[X]

If the Wiki is updated daily, I wonder why central game mechanic pages contain so much nonsense. Examples:

Skills:
http://crawl.chaosforge.org/index.php?title=Skills

The table listing the "Experience Required" is just plain wrong. According to the 0.10 source code this amount heavily depends on how many "skill points" the character already has acquired, and these skill points are somewhat related to experience spent (XP needed scales by a factor beginning at 1 and going up to 250). For example, if you first train a skill to 27 at -5 aptitude and then another skill to 27 at +5 aptitude, you usually pay much less XP for that than if you train the +5 skill first and then the -5 skill. The "amount of XP" needed to train a skill is not constant as the article suggests.

Weapon damage:
http://crawl.chaosforge.org/index.php?t ... pon_damage

This information is so bad it makes me cry. Almost nothing in the top level formula is correct for 0.10. The most important fact is not mentioned at all: base weapon damage is _multiplied_ by (1 + Weapon skill / 25) and then _multiplied_ by (1 + Fighting Skill / 30), i.e. at max. weapon and fighting skill the base damage of the weapon is multiplied by 3.952, while it's only multiplied by 2.233 if both skills are trained to 13.5. And the page says nothing about the fact that you can get a huge amount of extra damage by choosing a weapon with a strong bias towards either Dex or Str and having a large stat value. Example: An ogre with Str 30 and Dex 10 and maxed skills and no other bonuses will do 1d(20 * 116/78 * 2.08 * 1.9) = 1d(117.547) damage per hit with a giant club (100% strength weight), but only 1d(20 * 78/78 * 2.08 * 1.9) = 1d(79.42) with a blessed triple sword (60% strength weight).

Armour:
http://crawl.chaosforge.org/index.php?title=Armour

Huh? Armour evasion penalty slows down your attacks? According to the 0.10 source code, only shields lower attack speed, not armour.

Stabbing:
http://crawl.chaosforge.org/index.php?title=Stabbing

My, that table listing "stabbing multipliers" depending on skill, awareness and weapon type has very little to do with the actual game mechanics in 0.10.

[mageykun]

This information is so bad it makes me cry. Almost nothing in the top level formula is correct for 0.10. The most important fact is not mentioned at all: base weapon damage is _multiplied_ by (1 + Weapon skill / 25) and then _multiplied_ by (1 + Fighting Skill / 30), i.e. at max. weapon and fighting skill the base damage of the weapon is multiplied by 3.952, while it's only multiplied by 2.233 if both skills are trained to 13.5. And the page says nothing about the fact that you can get a huge amount of extra damage by choosing a weapon with a strong bias towards either Dex or Str and having a large stat value. Example: An ogre with Str 30 and Dex 10 and maxed skills and no other bonuses will do 1d(20 * 116/78 * 2.08 * 1.9) = 1d(117.547) damage per hit with a giant club (100% strength weight), but only 1d(20 * 78/78 * 2.08 * 1.9) = 1d(79.42) with a blessed triple sword (60% strength weight).

...there's also the matter of the 3 point difference in base damage. If you want to discuss the difference that str/dex weighting makes, wouldn't it make more sense to compare weapons with the same base attack? A GSC club by itself is always gonna be better than a triple sword, regarless of str. (You know, assuming you can equip it, and skills are right, etc).

Could I get some fact checking on this info? My working understanding has been that the first order considerations in damage output are base damage / delay, coupled with weapon skill since it multiplies base damage and improves delay. Fighting helps, but it's second order. Str / dex stacking is a second or third order, unless there's a lot of it.

[X]

Could I get some fact checking on this info?

Check the source code. The calculation is mostly in melee_attack.cc, and it's very, very complicated.

My working understanding has been that the first order considerations in damage output are base damage / delay, coupled with weapon skill since it multiplies base damage and improves delay. Fighting helps, but it's second order. Str / dex stacking is a second or third order, unless there's a lot of it.

I know that people think that way about damage output. But in fact (ignoring weapon speed for now) there is a factor up to 2.08 for weapon skill, up to 1.9 for fighting skill, and ranging between 1 (or even less for awfull stats) and 1.641 or more (for a weighter stat average of 36). These factors are not simply added but multiplied with each other, so there is a _huge_ synergy effect for maximizing all three factors. (And even more the resulting huge factor is multiplied with the speed gain factor of about 2, resulting in about 13 times the base damage of the weapon used without skill and stat bonuses).

...there's also the matter of the 3 point difference in base damage. If you want to discuss the difference that str/dex weighting makes, wouldn't it make more sense to compare weapons with the same base attack?

Both discussed weapons do 20 base damage, according to the wiki.

[mageykun]

Could I get some fact checking on this info?

Check the source code. The calculation is mostly in melee_attack.cc, and it's very, very complicated.

Why yes, thank you! I wasn't aware. Your condescension is most helpful. :p

Seriously, no disrespect intended- the source code is complicated, and people have misinterpreted it incorrectly more than once. I wasn't asking you to prove anything- you already presented your evidence and formulas. I'm just asking for a second opinion from someone else who's looked at the code, because that doesn't sound right to me (not that I'm always right or anything). Peer review is handy.

Both discussed weapons do 20 base damage, according to the wiki.

Bots and a quick 0.10 wizmode check: GSC 22, giant club 20, triple sword 19. Base stats are definitely one thing you can't trust the wiki to be up to date on, unfortunately (Maybe there's been a tweak in trunk, but I'd be surprised if the wiki reflected that before the bots).

[Galefury]

mageykun: He's comparing a giant club and a blessed triple sword, so base damage is 20 for both weapons.

X: The amount of exp needed to train a set of skills to specific levels is not supposed to strongly depend on the training order. You're saying it still does (even after some changes were made to prevent this), which means either you are misinterpreting the source or there is a bug.

Also, how to register an account on the wiki?

[tasonir]

To register an account you unfortunately have to send an email to register.crawlwiki@gmail.com, because of spam problems. Hopefully whoever monitors that (it isn't me) will get back to you quickly and you'll be ready to go.

Regarding weapon damage, look at the weapon he picked closely - Giant club (not spiked) and *blessed* triple sword, which are both 20

I have never heard anything about skill training becoming harder at higher levels - what source file is that in? I'm going to look up some of the damage formulas you mentioned, they sound pretty consistent with what I've heard, though. One question - does unarmed combat still get multiplied by UC skill or is the +1 base per level it?

Also, how does the 30/10 100% weighting come out to 116/78 and with 60% equal 78/78? Once I have these all straight I'll update the pages, have faith

[X]

X: The amount of exp needed to train a set of skills to specific levels is not supposed to strongly depend on the training order. You're saying it still does (even after some changes were made to prevent this), which means either you are misinterpreting the source or there is a bug.

There's no bug, all that code was very obviously changend intentionally, it is even documented in the sources. The 0.10 code works like this:

1. Skills need a certain amount of "skill points" to reach a given level. There is a table in the code listing the number of skill points. This cost is fixed. It does not depend on the aptitudes orthe experience level of the character.

2. The code determines a "character skill level" by adding skill points of all characters (function calc_total_skill_points in skills.cc). This sum is looked up in a table in skills2.cc (function skill_cost_needed) to determine the character skill leve". The table used is derived using some formula from this one (also in skills.cc):

0, 50, 150, 300, 500, 750, // 0-5
1050, 1400, 1800, 2250, 2800, // 6-10
3450, 4200, 5050, 6000, 7050, // 11-15
8200, 9450, 10800, 12300, 13950, // 16-20
15750, 17700, 19800, 22050, 24450, // 21-25
27000, 29750

3. Whenever fresh XP comes in, it is converted into a kind of "skill exercise unit". To the amount of XP needed for one excercise unit is taken from this table in calc_skill_cost() in skills.cc:

cost[] = { 1, 2, 3, 4, 5, // 1-5
7, 9, 12, 15, 18, // 6-10
28, 40, 56, 76, 100, // 11-15
130, 165, 195, 215, 230, // 16-20
240, 248, 250, 250, 250, // 21-25
250, 250 };

I.e. at the beginning of the game, at skill_level zero you just pay one XP per exercise unit, and over time it becomes more and more expensive, up to 250 XP per unit.

4. The exercise units are proportionally assigned to the skills in training (if there are not enough units for all skills, distribution is random).

5. One exercise unit yields ten skill points for that skill, modified by aptitude (i.e. ten at aptitude 0, twenty at aptitude +4, and only five at aptitude -4). This is in the function train_skills() in skills.cc.

6. The actual amount of skill points needed for a skill to reach a certain level is the value from the table

0, 50, 150, 300, 500, 750, // 0-5
1050, 1400, 1800, 2250, 2800, // 6-10
3450, 4200, 5050, 6000, 7050, // 11-15
8200, 9450, 10800, 12300, 13950, // 16-20
15750, 17700, 19800, 22050, 24450, // 21-25
27000, 29750

(i.e. the same table is used to derive the table for character skill levels as well as the skill points needed for a skill - this logic gave me a big headache to understand.)

--

And now back to proving that training order _does_matter: Let's assume we have two skills, one at aptitude +4 and one at -4. To maximize a skill, 29750 skill points have to be put into it (see table). How many XP does this cost?

A) Cost in Excercise units:

Aptitude -4: 29750 / (10 * aptitude factor) = 29750 / 20 = 1487.5 exercise units
Aptitude +4: 29750 / (10 * aptitude factor) = 29750 / 5 = 5950 exercise units

B) At overall character skill level ten, one training unit costs 18 XP. At skill level 27 one unit costs 250 XP.

C) XP cost depends on training order:

If you train the cheap skill at character skill level ten, and the expensive one at character skill level 27, you pay
(1487.5 * 18) + (5950 * 250) = 1514275 XP.

(Let's ignore that character skill level raises while you train a skill to keep the calculations simple.)

If you train the expensive skill at character skill level ten, and the cheap one at character skill level 27, you pay only
(5950 * 18) + (1487.5 * 250) = 478975 XP.

Of course this is an extremely unrealistic example, but the mechanism is there. If you want to train an expensive skill and a cheap skill, you're better off if you train the expensive skill first. However, this effect vanishes once you hit character skill level 23 because from then on all exercise units cost 250 XP.

--

Note: This description ignores some things that have an influence on skill cost, for example the discout you get for the stealth skill and others skills.

[X]

Also, how does the 30/10 100% weighting come out to 116/78 and with 60% equal 78/78? Once I have these all straight I'll update the pages, have faith

The formulas are:

1. Weighted Stat := (Str-Weight * Str) + ((1 - Str-Weight) * Dex)

(Although it's coded in a very funny way.)

2. If Weighted Stat exceeds 11:

Stat Damage Multiplier := (78 + 2 * (Weighted Stat - 11)) / 78

--

So, a weapon with 100% strength weight used by a character with Str 30 and Dex 10 yields:

Weighted Stat := (1 * 30) + (0 * 10) = 30
Stat Damage Multiplier := (78 + 2 * (30 - 11)) / 78 = 116/78.

A weapon with 60% strength weight on the same character yields:

Weighted Stat := (0.6 * 30) + (0.4 * 10) = 22
Stat Damage Multiplier := (78 + 2 * (22 - 11)) / 78 = 100/78.

The formula is in melee_attack::player_stat_modify_damage() in melee_attack.cc.

--

I just note that I accidentally used "78 / 78" in the original post instead of the "100 / 78" which should have been there, so the difference is less dramatic because both example weapons _are_ biased towards strength. But if you compare an evening start (80% Str) with a holy scourge (20% Str) you really get a _big_ difference.

[galehar]

Check the source code. The calculation is mostly in melee_attack.cc, and it's very, very complicated.

Indeed. Which is why you should check your facts before calling wolf.

1. Skills need a certain amount of "skill points" to reach a given level. There is a table in the code listing the number of skill points. This cost is fixed. It does not depend on the aptitudes orthe experience level of the character.

This is where you're wrong. Aptitudes have no effect on the number of skill points gained, they affect the number of skill points needed to reach a certain skill level. The order in which you train skills have no effect.

Note: This description ignores some things that have an influence on skill cost, for example the discout you get for the stealth skill and others skills.

This is fixed in trunk. Skill cost is now based on the total number of experience points earned. It's no longer influenced by the special cost of certain skills nor crosstraining.

It's true that the skill wiki page is confusing XP and skill points. However, since they are both hidden to the player, I'm not sure what good it would do to expose those complicated mechanism. The table is good enough to compare how expensive a skill level is compared to another. If we were to give more details, I think it would be better to just explain roughly the XP to skill points conversion mechanism (I guess that would be my job...). Giving away the actual conversion cost table would be an error. Players would think it's relevant, but it's really not. They have absolutely no control over skill cost, so they don't need to understand the details of it.

Weapon damage:
http://crawl.chaosforge.org/index.php?t ... pon_damage

This information is so bad it makes me cry.

I haven't bothered to check all your formulae, but I just want to point out that the effect of stats is minimal. Using your example in the fight simulator (30 str, 10 dex, maxxed skills, giant club vs blessed triple sword) the difference is 8.2%. And this is an extreme case.

Huh? Armour evasion penalty slows down your attacks?

True, this hasn't been the case for a very long time. Fixed.

[CommanderC]

@X:
I'm sorry to tell you that some of your calculations are wrong. The Stat Damage Base is not str_weight*str+(1-str_weight)*dex. It is str_weight*str+(1-str_weight)*(str+dex)/2.

The effect of Strength in melee damage isn't negligible, but the effect of Intelligence in spellpower is huge. If you plan to cast spells at some point in the game, raising INT is always the best choice.

[mageykun]

Blessing weapons still messes with the base stats? I thought we'd done away with that nonsense when we removed blessed katanas. I mean, it makes sense with the demon weapon transformations, but otherwise...

Back on topic, this seems like exactly the kind of thing that should not be on the wiki. Long, complicated calculations right from the source code are problematic. They're complicated, prone to misinterpretation, and can be downright misleading even when they're right.

[tasonir]

Yeah, I'm not likely to add exact calculations unless they are relatively clear and I'm certain of them. If the training of skills does slow down as you level up, then that should be mentioned, although I wouldn't give quantitative details. But galehar is claiming it doesn't, and even though I'm technically a programmer by trade, I appear to be a very bad one and have trouble reading the source ;P

The only thing which is exposed to me as a player that I thought was odd with skill training is that I can have two skills both at 0 apt, turn them both on from 0, and then kill some monster and one will be at .6 and the other at .3 Obviously the splitting isn't perfectly even when it doesn't round to an even amount of skill training units, but that was certainly not something I would have expected, which I now know a little bit more about. Otherwise this whole process isn't really shown.

[galehar]

If the training of skills does slow down as you level up, then that should be mentioned, although I wouldn't give quantitative details. But galehar is claiming it doesn't, and even though I'm technically a programmer by trade, I appear to be a very bad one and have trouble reading the source ;P

The cost of converting XP to skill points raises, so in some way, training does slow down. But it's balanced by the fact that XP is gained faster. I wasn't claiming that training doesn't slow, just that the order in which skills are trained has no effect. There's no way to optimize your skills by training first the ones with high aptitudes (or low).

The only thing which is exposed to me as a player that I thought was odd with skill training is that I can have two skills both at 0 apt, turn them both on from 0, and then kill some monster and one will be at .6 and the other at .3.

When you gain enough XP to train some skills but not all, the skill trained is chosen randomly (weighted by training values which is relevant for auto mode or focused skills).
If skill A and B are the only skill enabled, and skill A is trained first, there's no effort to train skill B the next time. We let the rules of statistics handle it You'll see some local variations, but overall, they will progress at the same speed.

[tasonir]

is the rate at which skill exp slows down based on your total skill exp, or character level? If it's based on your character level, then slow level races will have an easier time training skills, and 100-120 races would have less time to compensate for low aptitudes...

[galehar]

is the rate at which skill exp slows down based on your total skill exp, or character level? If it's based on your character level, then slow level races will have an easier time training skills, and 100-120 races would have less time to compensate for low aptitudes...

In 0.10, it is based on total skill points, which has some weird little side effects. Training cheap skills or using crosstraining makes skill cost increase a bit faster.
In trunk this is fixed, the skill cost is based on the total number of XP gained.

[tasonir]

Slaying bonuses are rolled seperately or in one roll? Ie, is it 1dFinal weapon damage + 1dSlaying, or 1dfinalWeapondamage+slaying? Are there any other additive slaying bonuses besides slaying, +dam on equipment, and enchantments on weapons? The 1.5 hand bonus, might, and racial bonuses are also mentioned.

I've updated http://crawl.chaosforge.org/index.php?t ... pon_damage but I just want to check that everything is correct.

Edit: Also added a small paragraph to skills explaining that it requires more exp to raise skills when you have high skills, and I read over armour but someone must have removed the part about bodyarmor slowing attack rate already, as I couldn't find it.

[tasonir]

I'd also like to make sure that unarmed combat (http://crawl.chaosforge.org/index.php?t ... med_Combat) is up to date, I was wondering about some of the information there. Does unarmed get mutliplied by weapon skill? ie, at level 27, do you get 30 base damage*(1 +27/25)*(1+fighting/30) or is it just 30 base damage*(1+fighting/30)?

Also, is "This attack delay is increased, however, by the EV penalty of your equipment. Every time you attack, the game rolls 1d10 + 2d(EV Penalty) - 10; if the result is positive, that amount is added to your attack delay." still correct? Armor isn't supposed to slow down attack speed for weapons anymore, is unarmed a special case? why? I assume this ev penalty is adjusted for armor skill?

What is the str/dex weighting of unarmed?

Edit: would love to pin down the exact base damage of statue form unarmed also - the knowledge bots list it as 3 + str /3 + UC; is that 3 the base damage of unarmed, or in addition to? ie, a non-claw character with 30 str would have 3 (base) + 3 (statue) + 10 (str/3) + UC? 26+skill, or just 23+skill? It's wrong on the unarmed combat page, and the statue form page assumes it's 23+skill.

[bobross419]

Not sure where else to mention this...

Kiku's page (http://crawl.chaosforge.org/index.php?t ... ubaaqudgha) says that in .10 you can't do the temple trick with abominations because Kiku turns all your undead helpers against you, but I'm not seeing that happen at all. Been about 30k turns now and running stable 0.10.2.

Keep up the great work on the wiki

[tasonir]

Based on just watching my delays while using unarmed and heavy armor, it seems like that formula is probably quite correct, it matches the non-scientific data I observed, so I'm fine with that. I can't seem to get damage reports working in explorer mode, it just said it was calculating average damage with <weapon> at skill level 1 through 27, but then I didn't see anything - where does the report go? is it written to a file? is it possible to get it to tell you your average damage per hit with a weapon in wizard mode, ranges, etc?

When I get a chance to test kiku abandonment in wizard mode I'll update the wiki.

Also, on an unrelated note, I have already updated the reaping page - http://crawl.chaosforge.org/index.php?title=Reaping - because I am fairly certain it is removed. It remains in the game on the two unrands I have listed. If this is not correct, please let me know, I'd hate to be updating a page with incorrect information. But I believe you cannot randomly generate reaping ammo anymore.

[mageykun]

Scythes of reaping might still be possible?

I can't find them explicitly removed in the change logs, but I also don't know how to batch create a mass pile of ego scythes in wizard mode to test (can't figure out how to get &o to create items with random egos. I could max polearms and spam Trog gifts, but then I have to sort the scythes out, which is too much work).

[Tiber]

I believe I've seen randart bows of reaping in games I played on 0.10 when it was trunk, and I don't recall seeing any change which affected that. I also believe that the chance of reaping working was changed to be based on damage done, not just on whether the reaping brand was used for the final blow.

Edit: here we go:
Dungeon Crawl Stone Soup version 0.10-a0-1282-g71db0c8 character file.
117687 | D:23 | Identified the cursed longbow of the Pill Bug {reap, rF+ rC+ MR} (You found it on level 23 of the Dungeon)

[onton]

Scythes of reaping might still be possible?

I can't find them explicitly removed in the change logs, but I also don't know how to batch create a mass pile of ego scythes in wizard mode to test (can't figure out how to get &o to create items with random egos. I could max polearms and spam Trog gifts, but then I have to sort the scythes out, which is too much work).

In a recent build (definetely not the latest one, probably two or three weeks old) I took off a scythe of reaping off Sigmund. Suites him well I guess

[galehar]

I can't seem to get damage reports working in explorer mode, it just said it was calculating average damage with <weapon> at skill level 1 through 27, but then I didn't see anything - where does the report go? is it written to a file? is it possible to get it to tell you your average damage per hit with a weapon in wizard mode, ranges, etc?

It makes a fight.stat file with the results of the simulation. In trunk, you also get the results in the message area (along with other improvements and... A documentation!).

[BlackSheep]

Scythes of reaping are still possible. Ranged weapons and ammo were removed a while back.

[XuaXua]

There's also the Sword of Zongoldruk, to a lesser extent.

[tasonir]

It makes a fight.stat file with the results of the simulation. In trunk, you also get the results in the message area (along with other improvements and... A documentation!).

For some reason I was looking for it in my morgues folder (before asking)...not sure why I figured it would be there. Found it now in the base crawl directory, which makes perfect sense

  Code:

Dungeon Crawl Stone Soup version 0.10.0

Combat simulation: Naga Fighter vs. worm (4000 rounds) (20120422/ 0:04:04)
Experience: 27
Strength  : 20
Intel.    : 19
Dexterity : 12
Base speed: 10

Monster   : worm
HD        : 5
AC        : 1
EV        : 5

Weapon    : +0 trident (vamp)
Skill     : Polearms

Skill | Bonus | Accuracy | Av.Dam | Av.HitDam | Eff.Dam | Max.Dam | Av.Time
  0   |  0.0  |  100%    |  12.87 |    12.87  |    9.27 |   472   |    13.9
  1   |  0.0  |  100%    |   9.87 |     9.87  |    7.40 |    24   |    13.3
  2   |  0.0  |  100%    |   9.76 |     9.76  |    7.59 |    21   |    12.9
  3   |  0.0  |  100%    |   9.35 |     9.35  |    7.76 |    23   |    12.0
  4   |  0.0  |  100%    |   9.14 |     9.14  |    7.92 |    24   |    11.5
  5   |  0.0  |  100%    |   8.90 |     8.90  |    8.05 |    23   |    11.1
  6   |  0.0  |  100%    |   8.65 |     8.65  |    8.44 |    21   |    10.2
  7   |  0.0  |  100%    |   8.59 |     8.59  |    8.82 |    22   |     9.7
  8   |  0.0  |  100%    |   8.23 |     8.23  |    8.90 |    23   |     9.3
  9   |  0.0  |  100%    |   8.26 |     8.26  |    9.70 |    21   |     8.5
 10   |  0.0  |  100%    |   8.02 |     8.02  |   10.02 |    20   |     8.0
-----------------------------------

Apparently 0 skill is the best way to go! Not sure what caused this error, but there was apparently a rogue calculation which caused one hit for 472 damage...Not sure why it cut off after level 10, either, the next report went all the way to 27 and didn't have this problem.

[galehar]

Apparently 0 skill is the best way to go!

I fixed a bug in trunk where skill 0 damage was boosted by distraction stabbing bonus. But it's not enough to do 472 damage! Anyway the fight simulator has been mostly rewritten in trunk so you should really switch to it if you want to play with fsim.

[cerebovssquire]

I did a quick writeup on DsFE earlier for tavern. It's on the wiki now because it is a lot less horrible than most stuff on the wiki: http://crawl.chaosforge.org/index.php?title=DsFE_guide
I deliberately kept it short (most of the chaosforge guides' problem is that they are long and detailed, tailored to the stuff the writers found on their characters).

P.S. Can I add "this guide is full of crap" to the headings of the bad guides (cough demonspawn conjurer cough vpie of kiku cough MDFi)
P.P.S. Can I add "this guide was written by a badass so think again if you are going to edit it" to my ones

The "difficulty of play" section for species is problematic. It's in the middle of the page, surrounded by objective data, but it's a subjective and often controversial section. Not sure what to do about it.

problematic = nonsensical and can we please kick it out? A little text stating the difficulties of individual species is OK in my book, but the scale is pretty useless/wrong.

  Code:

HOLY CRAP WIN! Found the randart "Duronadu", a +1 plate armor with RC+, RN+, MR, and a sad -4dam.

oh diary of a crawler, you make my day again and again

[banei]

oh diary of a crawler, you make my day again and again

The reason I wrote a diary was that I was under the impression it is like my own diary in my dresser. That is to say no one will ever read it. ever.

[cerebovssquire]

oh diary of a crawler, you make my day again and again

The reason I wrote a diary was that I was under the impression it is like my own diary in my dresser. That is to say no one will ever read it. ever.

I was really astonished upon finding out that not only had someone read my newbie diaries, but also liked them.
(they are hilariously horrible, though, from a strategy point of view)

I love reading other people's diaries IRL, I'm a bad person.

[cerebovssquire]

http://crawl.chaosforge.org/index.php?t ... MuWz_Guide

(it's better)

[galehar]

P.S. Can I add "this guide is full of crap" to the headings of the bad guides (cough demonspawn conjurer cough vpie of kiku cough MDFi)

If the guide is so bad and/or outdated, just delete it. The point of a wiki is that everyone can edit and improve it. Removing crap is improving. If the author disagrees and reverts the deletion, discussion an/or moderation can follow. Sure, it's possible that he is stubborn and insists that his crappy guide stays, and we won't fight him because we don't care enough. But my guess is that it's more likely that he won't care or he's not around anymore. So I'd suggest deletion.

P.P.S. Can I add "this guide was written by a badass so think again if you are going to edit it" to my ones

How about: at the time of writing this guide, the author had scored xx online wins. And a link to your CDO stats.

The "difficulty of play" section for species is problematic. It's in the middle of the page, surrounded by objective data, but it's a subjective and often controversial section. Not sure what to do about it.

problematic = nonsensical and can we please kick it out? A little text stating the difficulties of individual species is OK in my book, but the scale is pretty useless/wrong.

I agree. Maybe those advices would be more appropriate in a dedicated page in the strategy section. Similar to the existing Background and race combinations for beginners page. We could emphasise the subjectivity of those advices by calling the pages author's guide to choosing a powerful/easy species.
As I've already said before, such a page should make it clear if it's focusing on simple species (easy to play) or powerful species (easy to win).

[cerebovssquire]

How about: at the time of writing this guide, the author had scored xx online wins. And a link to your CDO stats.

I wasn't really being serious about that (definitely not about "badass" )... even the "the author had xx wins" gives the whole thing an arrogant tone I don't really want to have in something with a serious intention (guide) but maybe something like "written by X" with "X" being a link to X's stats is a good idea. It's just a matter of wording, of course.

Regarding the author's guide on choosing a species, we did recently have some top-end players commenting on their personal "ease of play" lists, so maybe that could go in. At any rate it's going to be more useful than the current ones, some of which are more "objectively wrong" than "right in a subjective context" or, at the very least. highly misleading. I remember minmay and mikee_ giving theirs and I think both were pretty reasonable.

I do think I am going to delete the MuWz guide of Kiku > Sif though. Even though I do sort of appreciate the work the author put into it, it recommends a clearly suboptimal strategy to achieve a goal similar to that my guide has, and a lot of work with bad results isn't worth all that much. Objections, anyone?