We are doomed..

On 2007-12-30 06:50:55, Gamer_2k4 <gamer2k4.DeleteThis@gmail.com> wrote:

> On Dec 29, 11:10 pm, "copx" wrote:
> > As I said: we are doomed!
>
> I was under the impression that Dwarf Fortress had already doomed
> us...

Coincidentally, you'll see an article by me discussing this exact issue on
http://gamesetwatch.com in exactly 57 minutes.

Andrew


--
The Roflwtfzomgbbq Quylthulg summons L33t Paladins -more-
"Apple: Celebrating the poisoning of Alan Turing since 1977."
ASCII Dreams: http://roguelikedeveloper.blogspot.com
Unangband: http://unangband.blogspot.com
 >> Stay informed about: We are doomed.. 

On Dec 30, 12:10 am, "copx" <c... DeleteThis @gazeta.pl> wrote:
> Thomas Biskup seems to be serious about finally releasing JADE in 2008 (seewww.adom.de).
>
> I cannot wait to try it. JADE is one of these mega-hyper-roguelike projects,
> which usually never get anywhere, but given TB's track record.. maybe he
> will actually do it..

I can wait. I've gotten quite good at waiting in the last n+1
years :>

> ADOM is already the most popular roguelike, and JADE is supposed to be at
> least a league above that, with a massive, living world and all that..

JADE, I worry, suffers from second-architecture syndrome. The second
system you architect is always the worst. The first system you don't
know better so slowly evolve a complicated ugly hairball that works
(ie, ADOM). The second system you try to plan ahead and build a
beautiful ivory tower that will handle any case cleanly and
efficiently. Unfortunately, said ivory tower usually lives only in
the clouds and doesn't actually work.

I eagerly await being proven wrong, however. IMHO, the best thing he
can do is get a 0.1 out without worrying about meeting the insane
expectations everyone has of him. My fear is that we're all building
the image of JADE being the perfect roguelike in version 0.1 (contrast
with ADOM 0.1, nothing like ADOM 1.0!) and so he feels the need to
keep it under wraps until it is good enough. Which is a sure recipe
for building into an ivory tower. The ugly foundations need to be
thrown to the wolves early to ensure they are strong enough to
withstand the features planned later.

> I guess for the rest of us that means we will have to take our projects and
> hide in shame
>
> Here is a copy of the JADE feature list, just to remind you:

I'll be joining Krice in the "I'll believe it when I play it" list.

To be honest, I'm more eager to hear of an ADOM update that fixes the
long standing bugs in the last version.

On an unrelated note, I'm having a lot of fun watching POWDER in
stress test mode. Last night I found another rare crash bug after
470,000 moves. Current run is up to 443,500 moves without crashing...
--
Jeff Lait
(POWDER: http://www.zincland.com/powder)
 >> Stay informed about: We are doomed.. 

"Gamer_2k4" <gamer2k4.RemoveThis@gmail.com> schrieb im Newsbeitrag
news:6ad8b499-35a8-4b9e-b748-99ce98ba0087@s8g2000prg.googlegroups.com...
> On Dec 29, 11:10 pm, "copx" <c....RemoveThis@gazeta.pl> wrote:
>> As I said: we are doomed!
>
> I was under the impression that Dwarf Fortress had already doomed
> us...

DF is not a roguelike. Full stop.
ASCII != roguelike. Back in my day all kinds of games used ASCII, bowling
sims, chess games, etc. Damn kids.
DF is a RTS game!

Yeah, there is the "adventure mode", but it sucks. It is nothing a real
roguelike developer has to fear. The silly combat system of DF makes some
people laugh, but that's the only 'good' think I have ever heard about the
"adventure mode".
 >> Stay informed about: We are doomed.. 

Am Sat, 29 Dec 2007 21:50:55 -0800 (PST) schrieb Gamer_2k4:

> On Dec 29, 11:10 pm, "copx" <c....DeleteThis@gazeta.pl> wrote:
>> As I said: we are doomed!
>
> I was under the impression that Dwarf Fortress had already doomed
> us...

With one difference... Jade shall be finished then, Dwarf Fortress will
not...

Bye...
Norbert
 >> Stay informed about: We are doomed.. 

On Dec 30, 10:04 am, Jeff Lait <torespondisfut... RemoveThis @hotmail.com> wrote:
> On an unrelated note, I'm having a lot of fun watching POWDER in
> stress test mode. Last night I found another rare crash bug after
> 470,000 moves. Current run is up to 443,500 moves without crashing...

Aren't those the worst, though? Crashes like that almost make me wish
I had coded my game in Java, so I'd know where (and what) the problem
was.

--
Gamer_2k4
 >> Stay informed about: We are doomed.. 

On Dec 30, 11:41 am, Martin Read <mpr....DeleteThis@chiark.greenend.org.uk>
wrote:
> Jeff Lait <torespondisfut....DeleteThis@hotmail.com> wrote:
> >To be honest, I'm more eager to hear of an ADOM update that fixes the
> >long standing bugs in the last version.
>
> ADOM Is Dead. Thomas no longer has any will to work on the code.

Please check http://www.adom.de/

He is planning a bugfix release for ADOM in 2008.
--
Jeff Lait
(POWDER: http://www.zincland.com/powder)
 >> Stay informed about: We are doomed.. 

On Dec 30, 5:04 pm, Gamer_2k4 <gamer....DeleteThis@gmail.com> wrote:
> On Dec 30, 10:04 am, Jeff Lait <torespondisfut....DeleteThis@hotmail.com> wrote:
>
> > On an unrelated note, I'm having a lot of fun watching POWDER in
> > stress test mode. Last night I found another rare crash bug after
> > 470,000 moves. Current run is up to 443,500 moves without crashing...
>
> Aren't those the worst, though? Crashes like that almost make me wish
> I had coded my game in Java, so I'd know where (and what) the problem
> was.

I'm not sure how Java would help. I'm running under it under the
debugger with full debugging turned on, so I get a very clean stack
trace when things go wrong.

That run, BTW, crashed at around 996,700 moves.
--
Jeff Lait
(POWDER: http://www.zincland.com/powder)
 >> Stay informed about: We are doomed.. 

Jeff Lait <torespondisfutile DeleteThis @hotmail.com> wrote:
>To be honest, I'm more eager to hear of an ADOM update that fixes the
>long standing bugs in the last version.

ADOM Is Dead. Thomas no longer has any will to work on the code.
--
\_\/_/ some girls wander by mistake into the mess that scalpels make
\ / are you the teachers of my heart? we teach old hearts to break
\/ --- Leonard Cohen, "Teachers"
 >> Stay informed about: We are doomed.. 

On Dec 30, 6:23 pm, Jeff Lait <torespondisfut....TakeThisOut@hotmail.com> wrote:
> I'm not sure how Java would help. I'm running under it under the
> debugger with full debugging turned on, so I get a very clean stack
> trace when things go wrong.

A debugger, eh? That might help; which one do you use?

--
Gamer_2k4
 >> Stay informed about: We are doomed.. 

On Dec 30 2007, 2:10 pm, "copx" <c....TakeThisOut@gazeta.pl> wrote:
> Thomas Biskup seems to be serious about finally releasing JADE in 2008 (seewww.adom.de).
> ...
> As I said: we are doomed!

The worst that can come of it is that we'll waste too much time
playing JADE to work on our own projects. One should not be
intimidated by feature lists, for two reasons- first, the proof is in
the gameplay not in the features. Second, several of the roguelikes
currently in development have feature lists which are just as
ambitious as JADE. I certainly hope that Mr.Biskup does capture
lightning in a bottle for a second time, but this kind of elevated
expectation probably won't do us or him any good.

- Joseph Hewitt
--
Comics> http://ataraxia.comicgenesis.com
GearHead> http://gearhead.roguelikedevelopment.org
GHForum> http://ghforum.chaosforge.org/
 >> Stay informed about: We are doomed.. 

Jeff Lait wrote:

> On an unrelated note, I'm having a lot of fun watching POWDER in
> stress test mode. Last night I found another rare crash bug after
> 470,000 moves. Current run is up to 443,500 moves without crashing...

On a related note, I thought your unrelated note was
interesting enough to start its own thread.

Do I understand correctly that you have a robot tester,
or "borg", that you can run in your game for the express
purpose of finding bugs?

Cool!

Now I'm imagining the feature set one wants for such a
thing: Like starting at a random level with a random
character after each crash (with some fraction exercising
the initial character-generation code), etc.

Do you proceed by feeding random characters into the
keyboard buffer, or have you coded an AI in a serious
attempt to win? Or some combination of the two, like
"every fifth move random" or something?

I made a standard debugging output where debug "trace"
statements write sentences about what the game is doing.
If the game crashes I can pop up TrDebug.txt and see
what it was doing right before it crashed. That has
helped a lot.

What do other people do?

Bear
 >> Stay informed about: We are doomed.. 

On 31 joulu, 18:45, Radomir 'The Sheep' Dopieralski
<n....RemoveThis@sheep.art.pl> wrote:
> The important question is why it's Java and not C++

I guess he has an advertising contract and is forced to
use Java.

> Time will tell, but I expect a large disappointment of the community --
> due to inflated expectations.

It's like the expectations to Duke Nukem Forever.
Only that Duke will exceed all expectations:)
 >> Stay informed about: We are doomed.. 

On Dec 31, 11:18 am, Ray Dillinger <b... DeleteThis @sonic.net> wrote:
> Jeff Lait wrote:
> > On an unrelated note, I'm having a lot of fun watching POWDER in
> > stress test mode. Last night I found another rare crash bug after
> > 470,000 moves. Current run is up to 443,500 moves without crashing...
>
> On a related note, I thought your unrelated note was
> interesting enough to start its own thread.
>
> Do I understand correctly that you have a robot tester,
> or "borg", that you can run in your game for the express
> purpose of finding bugs?

Yep.

> Cool!

Thanks, I highly recommend anyone with a mature roguelike to invest in
this sort of black-box testing. Roguelikes can have so many
interrelated conditions that it is nice to know they have been fully
tested.

> Now I'm imagining the feature set one wants for such a
> thing: Like starting at a random level with a random
> character after each crash (with some fraction exercising
> the initial character-generation code), etc.
>
> Do you proceed by feeding random characters into the
> keyboard buffer, or have you coded an AI in a serious
> attempt to win? Or some combination of the two, like
> "every fifth move random" or something?

The current system is pretty rudimentary. I create a random level
with a mostly-prebuilt character which reflects the sort of end-game
character who has lots of neat equipment. I then run it for 100 moves
using the standard creature AI to control the player. I then reset
and build another level.

This mostly stresses the level generation, which isn't that bad of an
idea with a roguelike. One tends to throw a lot of randomness at
level generation and it is one thing you want rock solid.

Because other creatures on the level are also active, there is a lot
of additional code paths being run as they run around and do various
things.

The AI to run the player didn't require any real new code - monsters
and the player are highly symmetrical in POWDER as a design choice to
make writing AI easier. My smart-monster AI is already built around
"what would a player do?" style thinking, so it already will test most
things if it is given a chance (ie, the creature has the right spells
and equipment, which my test character has)

One big limitation is there are many actions not exercised by the NPC
AI (they never read scrolls of remove curse or enchant weapon, for
example, so my crashing bug of 094 wouldn't have been found by the
stress tester).

My plan is to upgrade the stress tester in two ways:
1) Increase the length it will play one level up to 1000 turns from
100, provided the avatar is still alive. I'm pretty confident in my
dungeon generation now :>
2) Occasionally randomly create and use a potion, wand, scroll, etc.
This can ensure a wider net of actions will be covered.

> I made a standard debugging output where debug "trace"
> statements write sentences about what the game is doing.
> If the game crashes I can pop up TrDebug.txt and see
> what it was doing right before it crashed. That has
> helped a lot.

Yeah, I wish I did that with the message logs. The only log I have is
what the player would have seen as message text. So if the crash is
caused by an out-of-sight creature doing something fancy I don't have
the history text.

This run has got to 2,444,500 moves, so I think it is time I expand
the scope of the stress test.
--
Jeff Lait
(POWDER: http://www.zincland.com/powder)
 >> Stay informed about: We are doomed..