ANNOUNCE Eddie's patch V3.0.9e

On 2007-12-11 19:28:35, Eddie Grove <eddiegrove RemoveThis @hotmail.com> wrote:

> Kenneth 'Bessarion' Boyd writes:
>
> > On 2007-12-11 13:06:05, "Magnate" wrote:
> >
> > > Then why not make them instantly IDd the moment you know what they do? I
> > > thought that was your whole approach to reworking ID - as soon as the player
> > > knows, the item is IDd.
> >
> > Making that robust against external configuration, takes considerably more
> > effort.
>
> So long as we are going off on this tangent, there are all sorts of issues to
> resolve. If you are hit by a fire attack, does the player automatically know
> if s/he has fire resistance, or does s/he have to make a note? Current V does
> not tell you -- you have to watch your hp and guess.

I would say "automatically know if it's unequivocal". This ties into an
extensive revision I'm planning for a priori probability projections and
monster/player hp estimation of each other.

> If you do tell the player, do you tell which item if several are possible?

I'd say not worth a message, but augment psuedoid to track this sort of thing.
 >> Stay informed about: ANNOUNCE Eddie's patch V3.0.9e 

Eddie Grove wrote:
> "Magnate" <contact.me.TakeThisOut@some.other.way> writes:
>>>>> * Point-based character generation equivalent to best available from
>>>>> autoroller
>>>> Well done! Max total starting stats is 85, is that right? Point-based
>>>> clearly the way to go now.
>>> Not so simple. Non-mentioned start at 10.
>>>
>>> 2 @ 18
>>> or
>>> 3 @ 17
>>> or
>>> 4 @ 16
>>> or
>>> 6 @ 14
>>> Unfortunately not 5 @ 15. 4 @ 15 + 1 @ 14. Oh well.
>>>
>>> You can get 3 @ 17 with the autoroller if you are patient. That was my
>>> base.
>>> I don't particularly care what the choices are so long as tedium with the
>>> autoroller gives no advantage.
>> Well, what you have there is not at all like the autoroller. The autoroller
>> has only one limitation - the hard-coded limit on the total of all six
>> starting stats (which I think is 84 but cannot check now). Actually no, the
>> autoroller cannot give stats higher than 17 or lower than 8, so that's
>> another limitation.
>
> What I did is not perfect, but is an excellent approximation. The old
> point-based allowed 18's, at a terrible price. I kept that, but made the
> price a smidgeon less. I don't really care, but I was trying to combine both.
>
> I have never seen even a 14 along with 3 17's from the autoroller. Maybe
> never a 13. I will take you at your word, and then ignore you. I like what I
> did. It keeps the flavor of old point-based at an equivalent power level to
> what I have experienced from the auto-roller.
>
> I don't mind being overruled, if that's what it takes for Andrew to dump the
> autoroller, but I feel that I did a good job of merging the two approaches.

Sorry, I misunderstood. When you wrote "Point-based character generation
equivalent to best available from autoroller" I thought you meant
theoretically available. I merely pointed out that you have not exactly
duplicated the autoroller's theoretical capability, but that's not a
problem. I have no problem with you approximating it to your perception
of the autoroller's observed best, since that's probably equivalent to
what the autoroller gives about 99% of the time, and is easily enough to
argue that there is no longer any point keeping the autoroller - which
is your aim.

>> Yeah I know what autoinscriptions are (one of NPP's best innovations), but I
>> didn't understand your point about applying them to items learned by
>> testing. What's the difference between that, and applying them to items IDd
>> by magic?
>
> Say you use an unaware rod of trap location, and thus learn awareness. Then
> it ought to be autoinscribed. Dropping it and then picking it back up to get
> the inscription is clunky.

Ah, I didn't know that autoinscription had that clunkiness. Well done
for fixing it.

>> That's fine. Did you try the alternative randart.c I linked for you? Not
>
> I usually play NPP 0.4.1 for randarts. Is it much different?

I don't know - I can't remember whether Jeff adopted parts of that code
before or after 0.4.1, and I don't know which version of NPP behaves
most like it.

> In any case, I am suggesting changes to V, and restricting myself to things I
> have not seen elsewhere or can't stand to play without. Andrew is perfectly
> capable of getting your version of randart.c without my help.

I know he is - I wasn't lobbying for it to be included in V, I was
hoping to get your views on it. You usually have plenty to say about
randarts and I would be interested to know if that code generates
randarts you consider to be better (more interesting/useful rather than
more powerful). Having said that, both Jeff and UnAndrew have
significantly improved upon it, and when Takkaria does rewrite V's
randart.c he should look to theirs rather than that one. So while I
would be interested in your views on it, your views on the randarts in
the newest Un and NPP would probably be more important.

>>> This is useful if you know aggravation or teleportation, but technically
>>> have
>>> not learned the flavor. I hate having to take notes on "this flavor is
>>> that".
>> Then why not make them instantly IDd the moment you know what they do? I
>> thought that was your whole approach to reworking ID - as soon as the player
>> knows, the item is IDd.
>
> That is complicated and involves redoing an enormous amount of stuff. I think
> it is too radical to propose incorporating to V the way I implemented it.
>
> I do include auto-id on things like a ring of con when you wield it.
> Keeping track of when you take less damage from a fire attack is different.
>
> The way I did things when I was considering doing a variant, once you learn a
> property, you recognize it on future items, so if you figure out a sword of
> flame you immediately [perhaps modulo pseudo-id] recognize the next mace of
> flame, and that randart axe might be Axe(2d8)(+9,+9) {flamebrand ? ?} to show
> three properties one of which is flame and 2 that you do not know.

Interesting. IMO you either need magical ID or you don't. If you don't,
then you eventually find out everything about an item by a combination
of use and pseudo - as you say, this is complicated and involves
reworking lots of stuff. If you do need magical ID, then I don't rate
making it slightly less important as a very high priority. But I'll try
your patch and see if I come to revise that opinion!

Cheers,

CC
 >> Stay informed about: ANNOUNCE Eddie's patch V3.0.9e 

Magnate <chrisc RemoveThis @dbass.demon.co.uk> writes:

> Eddie Grove wrote:

> > I do include auto-id on things like a ring of con when you wield it.

> Interesting. IMO you either need magical ID or you don't.

This is not about what is needful, but what annoys me!

If I put on the ring, and it boosts my con by 3, it really irks me to open a
notes file and write down that sfjl khkwe sdh means con. Or when I wield
multiple aware rings of con and I have to inscribe them to keep track of which
is which. One should not need ?id to keep track of obvious details.

It's a small step in the right direction, and I don't see any rational
argument against it. Perhaps after a hundred small steps people will get used
to the idea of trying things out to see what they do.

I've been playing a lot of "no selling to stores" but without increased gold
drops. In those games, as a diver, I cannot afford niceties like lots of ?id
until 2000' or so. This is probably a better game, but too radical for V.

So long as players can survive, less gold means more interesting decisions.
However, I intended to err on the side of more than enough gold to keep the
game playable for beginners. People will have to tell me if they think gold
drops need to be changed in either direction. Currently at 200' and below
money drop values are multiplied by a factor of 4.

So long as I am rambling, let me add that one of the benefits of better
starting stats is that it is less painful to have to quaff that potion of
weakness in the early going rather than selling it for awareness.


Eddie
 >> Stay informed about: ANNOUNCE Eddie's patch V3.0.9e 

On 2007-12-10 23:49:06, Eddie Grove <eddiegrove.DeleteThis@hotmail.com> wrote:

> > > will_asher writes:
> > >
> > > > Man, I really like everything you've done in this patch except for these two:
> > > >
> > > But before you go changing what I did, why not give it a sincere try?
> > > Perhaps you will learn to appreciate why I did it.
> >
> > I will give it a couple tries first before I change it, of course.
> >
> > Would it be better if I wait till I wait for a newer version of the patch before
> > putting it in my semi-variant?
>
> Yes, I think so. Many code fragments are embarassing to me.
> They hopefully work correctly, but are not easily modifiable and extendable.
>
> I wanted to get it out so people can try things and eventually discuss what
> they would like to see make it into official V in say 3.2.0 next summer.
> I did not include my truly radical preferences for that reason.

It turns out that I like it even with not being able to sell to shops. I was
worried about having a shortage of gold, but a actually kindof think you
overcompensated a little with the gold drops.
 >> Stay informed about: ANNOUNCE Eddie's patch V3.0.9e 

will_asher <will_asher RemoveThis @yahoo.com> writes:

> It turns out that I like it even with not being able to sell to shops. I was
> worried about having a shortage of gold, but a actually kindof think you
> overcompensated a little with the gold drops.

That could be. I'd love to get feedback on this. I paid careful attention
over several games to how much money I got by selling compared to how much I
brought home, and it was definitely more than the factor of 4 I used in the
patch, but I don't know how much that is dependent upon my personal playstyle.


Eddie
 >> Stay informed about: ANNOUNCE Eddie's patch V3.0.9e 

On 2007-12-12, Eddie Grove <eddiegrove RemoveThis @hotmail.com> wrote:
> Magnate <chrisc RemoveThis @dbass.demon.co.uk> writes:
>> Eddie Grove wrote:
>> > I do include auto-id on things like a ring of con when you wield it.
>> Interesting. IMO you either need magical ID or you don't.
> This is not about what is needful, but what annoys me!
>
> If I put on the ring, and it boosts my con by 3, it really irks me to open a
> notes file and write down that sfjl khkwe sdh means con. Or when I wield
> multiple aware rings of con and I have to inscribe them to keep track of which
> is which. One should not need ?id to keep track of obvious details.

I guess the philosophical argument is that this relies on the player's knowledge
rather than the character's. You might know that there is only one type of ring
which can boost CON, your character doesn't.
 >> Stay informed about: ANNOUNCE Eddie's patch V3.0.9e 

On 2007-12-17 17:38:40, Paul Murray <paul DeleteThis @murray.net> wrote:

> On 2007-12-12, Eddie Grove wrote:
> > Magnate writes:
> >> Eddie Grove wrote:
> >> > I do include auto-id on things like a ring of con when you wield it.
> >> Interesting. IMO you either need magical ID or you don't.
> > This is not about what is needful, but what annoys me!
> >
> > If I put on the ring, and it boosts my con by 3, it really irks me to open a
> > notes file and write down that sfjl khkwe sdh means con. Or when I wield
> > multiple aware rings of con and I have to inscribe them to keep track of which
> > is which. One should not need ?id to keep track of obvious details.
>
> I guess the philosophical argument is that this relies on the player's knowledge
> rather than the character's. You might know that there is only one type of ring
> which can boost CON, your character doesn't.

From a background knowledge point of view, it could be argued that the character
should have a *better* understanding of the game world than the player (at least
when both INT and WIS are high).

As another example, exposing exact damage roll values is currently a wizmode
debugging feature. But in-game, the character clearly knows (qualitatively) how
hard the blows (s)he is making, are; the coding-simplest way of translating this
qualitative information into a player-usable form would be to report the hp
damage. (Adjectival translation is harder, both technically and keeping U.S.
"family-safe".)

More generally, "monsters cheat" AI option wouldn't be cheating if the player
had the same feedback the AI does. At that point, the only reason not to
default it on, would be the self-healer problem.

This, of course, would be a reversal of a design decision made back in Moria.
None of this is V-suitable.
 >> Stay informed about: ANNOUNCE Eddie's patch V3.0.9e 

On 2007-12-17, Kenneth 'Bessarion' Boyd <zaimoni DeleteThis @zaimoni.com> wrote:
> On 2007-12-17 17:38:40, Paul Murray <paul DeleteThis @murray.net> wrote:
>> On 2007-12-12, Eddie Grove wrote:
>> > Magnate writes:
>> >> Eddie Grove wrote:
>> >> > I do include auto-id on things like a ring of con when you wield it.
>> >> Interesting. IMO you either need magical ID or you don't.
>> > This is not about what is needful, but what annoys me!
>> > If I put on the ring, and it boosts my con by 3, it really irks me to open a
>> > notes file and write down that sfjl khkwe sdh means con. Or when I wield
>> > multiple aware rings of con and I have to inscribe them to keep track of which
>> > is which. One should not need ?id to keep track of obvious details.
>> I guess the philosophical argument is that this relies on the player's knowledge
>> rather than the character's. You might know that there is only one type of ring
>> which can boost CON, your character doesn't.
> From a background knowledge point of view, it could be argued that the character
> should have a *better* understanding of the game world than the player (at least
> when both INT and WIS are high).

However this examples relies on having knowledge of all possible types of object,
which the player can have by reading the data files, but the character cannot.

If you are going to auto-id when an effect is unique, it would be a better idea
to scan the object list to determine that the effect is unique for that object type,
rather than hardcoding it, that way it will be robust against data file changes.

Either way though, you are assuming a world knowledge that the character cannot have.
(Still probably a good idea from a game enjoyment point of view, as the player
*does* know.)

With the no selling to stores, an alternative to boosting gold drops would be to have
squelching convert the object into gold, at some discount from selling price. This
would maintain the gold income while eliminating the frequent returns to town to sell
loot, but is hard to justify in game world terms.
 >> Stay informed about: ANNOUNCE Eddie's patch V3.0.9e 

I noticed that the store interface no longer shows help text at the
bottom of the screen. This was somewhat confusing since many variants
use "x" to examine an item in the store, but vanilla uses "l". I
think the help text should be added back in.
 >> Stay informed about: ANNOUNCE Eddie's patch V3.0.9e 

On Dec 18, 11:26 am, Kenneth 'Bessarion' Boyd <zaim....RemoveThis@zaimoni.com>
wrote:
> On 2007-12-18 10:50:40, Paul Murray <p....RemoveThis@murray.net> wrote:
>
> > On 2007-12-17, Kenneth 'Bessarion' Boyd wrote:
> > > From a background knowledge point of view, it could be argued that the character
> > > should have a *better* understanding of the game world than the player (at least
> > > when both INT and WIS are high).
>
> > However this examples relies on having knowledge of all possible types of object,
> > which the player can have by reading the data files, but the character cannot.
>
> That is a choice of background. It's not that the "character cannot", it's that
> an intentional choice was made to make the character not have enough legendary
> knowledge. That design choice can be changed; it's just too radical a change
> for V.

There is essentially no difference in a game with the standard
artifacts *ID*ed automatically or not. Anyone reasonably experienced
in the game knows them anyway; the main reason to *ID* them is often
to get a pretty character dump, not for any gameplay-related purpose.
I know that some people like to avoid being 'spoiled' about the
artifacts, but I do wonder just how much they actually care.


> Using INT and WIS as proxies for that legendary knowledge can be rationalized,
> once the decision to allow the character that knowledge at all is rationalized.
>
> > If you are going to auto-id when an effect is unique, it would be a better idea
> > to scan the object list to determine that the effect is unique for that object type,
> > rather than hardcoding it, that way it will be robust against data file changes.
>
> Exactly. More difficult to implement, but it's the lowest-future maintenance
> way of doing it.
>
> However, once that policy decision is made there is a strong temptation to go
> further and implement partial auto-id.

UnAngband already has this, to a degree.
 >> Stay informed about: ANNOUNCE Eddie's patch V3.0.9e 

Roger <rogercnorris.RemoveThis@yahoo.com> writes:

> I noticed that the store interface no longer shows help text at the
> bottom of the screen. This was somewhat confusing since many variants
> use "x" to examine an item in the store, but vanilla uses "l". I
> think the help text should be added back in.

I could not figure out how to add a new command character in the stores, so I
had to snag an existing one. I took the '?' for the help command. Thanks in
advance to anyone who is willing to fix it.


Eddie
 >> Stay informed about: ANNOUNCE Eddie's patch V3.0.9e 

On 2007-12-18 10:50:40, Paul Murray <paul.DeleteThis@murray.net> wrote:

> On 2007-12-17, Kenneth 'Bessarion' Boyd wrote:

> > From a background knowledge point of view, it could be argued that the character
> > should have a *better* understanding of the game world than the player (at least
> > when both INT and WIS are high).
>
> However this examples relies on having knowledge of all possible types of object,
> which the player can have by reading the data files, but the character cannot.

That is a choice of background. It's not that the "character cannot", it's that
an intentional choice was made to make the character not have enough legendary
knowledge. That design choice can be changed; it's just too radical a change
for V.

Using INT and WIS as proxies for that legendary knowledge can be rationalized,
once the decision to allow the character that knowledge at all is rationalized.

> If you are going to auto-id when an effect is unique, it would be a better idea
> to scan the object list to determine that the effect is unique for that object type,
> rather than hardcoding it, that way it will be robust against data file changes.

Exactly. More difficult to implement, but it's the lowest-future maintenance
way of doing it.

However, once that policy decision is made there is a strong temptation to go
further and implement partial auto-id.
 >> Stay informed about: ANNOUNCE Eddie's patch V3.0.9e 

On 2007-12-07 19:55:21, Eddie Grove <eddiegrove.TakeThisOut@hotmail.com> wrote:

> I've put together a bunch of changes to V that I hope improve the game.
> My underlying goals are to improve squelch, remove any need for taking notes,
> and to reduce the impact of identify. I've also included a few requests.
or sale

I've been trying out your game... I was a bit skeptical about the lack of
selling, but it seems to work well. You can see my current dump here:
http://angband.oook.cz/ladder-show.php?id=7340. Character was basically a true
power-diver until around 2300'- I'm afraid I slowed down a bit there though,
after a near-fatal surprise encounter with Adunaphel with less than 300 hp. Then
went very fast from 2500 to 3500, where I am now. Recently finished with !Con,
but am in a quandary as to which rings to wear- will probably soon find a big
enough Speed ring to moot the question, but...

Anyway, thought you might want to see someone playing your patch... next time i
will try for a pure diver.
 >> Stay informed about: ANNOUNCE Eddie's patch V3.0.9e