All Projects

I've always been assuming that higher stats will make learning skills easier.
The bars on my skill panel tell otherwise.

Here's the raw data. I'm listing widths of green bars, in pixels. (The two skills with a star have blue-yellow bars, where I'm listing the yellow bar length - blue bar was generally over the full bar length.)
The overall width of all bars is 78 pixels, excluding borders, on the unmodified Elves theme.

The various columns are for before and after the listed stat trainings.

The above numbers show that different stats have different effects on each skill, which I didn't expect but it does make sense.
What does not make sense is that an improved CHA makes Brewing harder.

Increasing the green bar of a skill by one pixel used to take as many "You gained some experience with <insert skill here>" messages as having the skill bar switch from "full green" to "blue-yellow-green".

Since the introduction of higher stats, I have been observing that the switch takes an extremely long time - at least four times the number of messages required to increase the bar by a single pixel.

I'm tentatively classifying this as a bug (it could be intended behaviour, in which case please change this to "feature request").

My first wild guess would be that this could be that the green bar is drawn a few pixels too wide, and then clipped when displayed in the Stats&Skills windows; the green bar would then increase normally as it should, but we'd simply not see the last few pixels.

The player is sent to fetch ink and paper from Jayose, but the corresponding quest menu item is missing.

Workaround:
Typing “give me ink and paper” in NPC chat while targetting Jayose does still work.

(Not marking this confidential as no useful information is revealed to those who don’t know the quest, and no new information is revealed to those who know it.)

In the Interview Sunshine Squadron Applicants quest, the menu choice on Zak reads “I was sent by Raylour”.
That a in Rylour’s name is wrong.

My book of stocks says that I can heat heated platinum stock/ingots to super-heated, then red-hot, then quench.
It does not allow me to smelt platinum ore, nor heat unheated platinum stock/ingots, nor hammer them.

I’m not sure how much of that is intentional.
It’s strange that I can’t heat unheated stock/ingot though.

When I return to Vladovic with the Maajo, I’m offered a “Hi again” option, but he does not understand it.
Giving him the Maajo is met with “he does not need that”.

It didn’t occur to me to try the other quest item first.
I guess getting the equivalent of a blank stare at an explicitly offered quest answer qualifies as a bug anyway

Sometimes, the client thinks I pressed the Shift key, but does not notice when I release it.
(This happens regularly when I shift-click in the inventory, for example.)

This is on Ubuntu Linux, with German keybindings.

May be related to PS#3070.

N.B.: I cannot press “Shift” when defining shortcuts. Pressing a shift key will give me “LShift” or “RShift”.

https://planeshift.svn.sf.net/svnroot/planeshift/trunk/docs/compiling.html lists the link to Jam as ftp://ftp.perforce.com/pub/jam/ .
The directory has moved, the link is now ftp://ftp.perforce.com/jam/ (without the /pub/ ).

As an aside note, it would be nice if there were a hint about which version of Jam has been tested to work with the PS sources.

psupdater alternates between two results for me:

1. Failure

Output is

    PlaneShift Updater Version 1.10 for linux64.

    Checking for updates:

then a hang that I usually kill via Ctrl-C.

2. Success

Output is

    PlaneShift Updater Version 1.10 for linux64.

    Checking for updates:
    Checking for updates to the updater: No updates needed!
    Checking for updates to all files: No updates needed!

    Updater finished, press enter to exit.

psupdater alternates between these two behaviours, occasionally having two fails or two successes in a row.
The first run after reboot/login usually fails.

This is not a serious problem for me as I can always restart psupdater.
However, such behaviour Should Not Happen, which means something is seriously wrong somewhere near the networking layer (and I’m a bit worried that this might affect the game itself).

Nature intuition is far too precise; getting just a dump of player data isn’t very IC.
Also, in practice, many people object to being scanned, mostly because they’re unsure what information exactly the mage got and what he can do with it.

I can imagine several approaches to address one or both of these points:

1. Don’t give all stats and skills, just a random selection. Higher spell levels might return a larger selection; e.g. 1 stat/skill at AW 10-19, 2 at 20-29, etc.

1a. To make the spell work better RP-wise and technically in consensual situations, the target can actively select which of its stats/skills the mage should get.

2. Inform the target that it is getting scanned, with a message like “Yannae is looking into your soul and takes a glimpse at your Strength and Agility”.

3. Make the information imprecise. E.g. if the true value is 100, tell the mage it is “between 80 and 110” (rarely, tell him it’s “between 60 and 90”). Make sure this scales properly - if the result from a skill value of 100 is “80..110”, then the equivalent result for a skill value of 10 should be “8..11”, not “-10..20”. (The math to get all borderline cases right is a bit tricky, I’ll spill all the gory details on request.)

This is related to  FS#646  but not the same.
Feel free to reduce the severity if you think it’s overrated. I’m setting it High because it adversely affects Roleplay.

 FS#646  had comments that Nature Intuition should give just the stats, not the skills.
I disagree - a skill is part of a person’s mental landscape just as his willpower, for example.

I’m missing the following mobs, about 30 minutes into game time after a server restart:

* Marfusang & trepor near the BD side of the imago consumer valley of bdroad2
* All except two imago consumers in the valley itself
* Trepor herd and warrior trepor in bdroad2
* Marfusang pair between the three trees further down the road from BD
* Marfusangs and Vilenaut lurking behind the boulders in BD area
* The Dark Circle Rogues in BD area.
(I haven’t checked the gold mine ulbernauts in bdroad2, nor did I check bdroad1.)

The first mobs (marfusang & trepor) I just killed myself, and they didn’t respawn within 30 minutes or so (maybe even longer).
I don’t know whether the cause of the other mobs is the same or something unrelated.
Curiously enough, the two remaining imago consumers will respawn just as quickly as mobs elsewhere do, so my knee-jerk speculation that this might somehow be related to the map itself is probably wrong.

Also, my /pos near the marfusang/trepor positions gives “Your current position is bdroad2 374.52 34.06 145.95, instance: -108130976”.
That instance number looks suspicious to me, though I *have* met one other player running through it.

Here’s the procedure:

Attack a mob.
Run away from it, and cease attacking (click the white flag).
Log out before it reaches you (this is crucial, the monster must still be running at the time when you log out).

Mob continues running, and vanish after a while.
Onlookers continue to see it running and getting held up by obstacles.
However, this seems not to be what’s happening on the server: trying to target the mob will give you a “Mob is out of range”-style error message. Also, if the player logging off logs on again, he won’t be able to see the mob.
After a while (two to four minutes), the mob will vanish even for the observer.

Untested variations:
* We haven’t tested whether the attacker’s state (attack ceased or in attack mode) influences this bug.

<speculative mode>
The mob simply continues to run, disregarding any obstacles; the server doesn’t see any reason to send updates to the client because the mob state doesn’t change. Clients extrapolate the mob being held up by obstacles, but targetting contacts the server and gets back the “mob is too far” message.
Once the mob falls off the edge of the world, the mob is taken out of the game and observers’ clients are notified of that change, at which point the image of the mob is removed from screen.
This is something that I’d liked to have tested, but I haven’t found a way to verify that. (It would require an observer who’s so far that he can’t see the mob, but placed precisely enough on the mob’s escape route to see it when it passes. A GM might be able to place the attacker and the observer on known coordinates.)
</speculative mode>

Sack is 30, which seems OK to me.

Crates are 20, which is less than a sack. Judging by appearance, 3x3x3 sacks would be roughly the same volume as one crate, so a crate should have 30x3x3x3=810 capacity.

Two-door closets are capacity 50. Given appearance, they should offer even more volume than crates.

Glyph sack are 20, could be 16 (16 glyphs @ size 1.0 each).
A nice though not overly important side effect would be that this would make “spilling” into the glyph sack slightly less probable.

(Fixed containers like kitchen table, furnace, forge, and stock casting seem to be just fine.)

Why I want this: I’m tasked with reducing lag inside a guildhouse. Sacks and crates would be ideal since putting stuff in them would reduce the number of objects in the house. Sacks would be the best option, but we’d need dozens of sacks just for the armory, and it would become difficult to keep track of what’s in which sack.)

Reason 1: Items in the shelf remain where they are, which looks silly.
Reason 2: You simply don’t pick up a shelf while there’s stuff in it, do you?
Reason 3: It’s hard enough to place a shelf so it’s aligned with the house wall. It’s easy to accidentally pick up the shelf instead of, say, the blade in it. This combination is making those unhappy who placed the shelf (and earns the bumbling idiot who picked up the shelf a serious telling off).

Possible implementation strategy: whenever an item is placed so that it is supported by another item (by placing it on top, or in the case of shelves, on a board), put the id of the supporting item in a “supporter” field of the newly placed item.
When picking up an item, the server could find out what items would lose support, and respond appropriately:
1. If the player picks up just that item, the items that rest on it fall down. (This would also make the old fly exploit impossible, so crates could be allowed as makeshift tables again.) Falling down could be handled just like a character who’s falling from a cliff.
1a. If a dev is feeling lucky, he can add fragility factors to all items in game and let stuff break if it falls too far. (That would make players unhappy. Adding fragility factors to every movable item would also be a lot of work.)
2. If the item plus the stuff resting it is more than the character can carry, the item won’t budge. Player would get message “You can’t move this item out from under all the stuff that is resting on it.”

You need to limit how much can be stacked on each other, otherwise people will start building Towers of Babylon with crates.
Either limit the number of items that can be stacked on top of each other. (Limited, unrealistic, but relatively simple to implement.)
Or place a weight limit that each item can be loaded with, and have it break if that limit is exceeded. (Realistic, plus it will give new interesting scenario possibilities, such as the bridge that will break if the character carries more than leather armor and two weapons. On the minus side, implementing all the ramifications will add several weeks of developer work.)

The message is misleading. Here’s what the source code has to say:

// If the slot next attack time is not yet up, abort (another event sequence should have been started)

The real problem, however, is that this message (and the subsequent combat abort) are happening spontaneously while in mid-fight. This means one participant stops combat while the other continues slashing happily.
I have seen this happen both to players and NPCs (I attacked an ulbernaut, had to run because I got the message, after a minute of happy chasing the ulbernaut lost interest, too).
I think this has been happening since 4.01, while the message was added in one of the more recent server updates (svn blame attributes it to revision 2077). I suspect this is a duplicate of bug reports a la “NPCs and player characters spontaneously lose interest in combat”, but I haven’t found them - doubtlessly because I’m searching for the wrong keywords.

Personally, I think it’s Bad Design(TM) to assume that just because I’m timing out, some faraway code must have started another event sequence.

Improvement proposal follows;
heed, modify, or ignore as appropriate,
as I’m certainly not aware of all ramifications:

It would be better to have an object that holds a representation of the currently-active activity, with the ability to abort that activity (activities would need a virtual function to accept such an abort, and return a message for the player).
E.g. the Dig action has a Start Message (”You start digging”) and an Abort Message (”You stop digging”); the abort message isn’t sent directly to the player but returned to the caller of the new virtual Abort() function.

With that architecture, PS can even implement a general display policy for abort messages: since the current-activity object gets the message, it can display it as appropriate. E.g. it could always make the abort message appear in red in the middle of the screen so players are generally alerted to activity aborts (as it is now, it’s very easy to accidentally abort digging without even noticing it).

There’s a valley near ojaroad1/-330/35/589 that’s impossible to escape because the slopes are not climbable.

Me and Yanlora Kithbeth got trapped in it today.

Suggested fix: wall it off so nobody can get into it (the valley opens into the Big Blue anyway).

Whenever I try to leave Laanx dungeon, I get “Received out of bounds positional data, resetting your position” and I get reset to the cellar near the normal exit.
This happens regardless of whether I’m exiting through the cellar or go to the top.

Also, when I peer through the exit to where the free sky should be, I just see areas that don’t do graphic refresh.

All leather armor (boots to torso)
All chainmail armor (boots to torso) (not 100% sure, I think I saw this during buying/selling)
All plate armor (boots to torso)
Kartoffel (Reported for 0.4.00 by Mario Rohkrämer (LigH) in #1233, confirmed for 0.4.01)
Snowbud (Reported by Mario Rohkrämer (LigH) in #1233, not confirmed yet)

Seems like one icon artist in the Art team hasn't found out how to properly set transparency yet.
Nice job on the icons themselves though

This is a suggestion to make it easier to announce the date and time for events such as tournaments, roleplay events, etc.

The date and time is often given as a real-world date/time such as “on Sunday, May 25th 15:00 PST”, which is
(a) difficult to translate to the local time zone for most people and
(b) a real-life reference that damages immersion.

I suggest a conversion tool to remedy this: a window with two selection tools for datetime, one for local datetime, one for Ylliakum datetime. Selecting a datetime in Ylliakum sets the local datetime and vice versa.

When writing announcements, players can select a local time and see the Ylliakum time, ready for typing into the announcement.
When reading announcements, players can select Ylliakum time and see the datetime of the event in their local timezone so they can plan their real life around it.

The selection tool would be a calender and a clock. The clock’s resolution would have to match the resolution of Ylliakum time (i.e. 3 AM, 6 AM, 9 AM, 12 AM, 3 PM, 6 PM, 9 PM, 12 PM).

Future dates/times would depend on how stable Ylliakum date/time is. I.e. after a server crash, Ylliakum time would have to be derived from real time instead of simply continuing from where it stopped (I don’t know whether this is done already, or if the server has any idea of a current date at all).
Alternatively, one could extend the text output routines to accept control codes for Ylliakum time, e.g. [2008-05-24 9:10] anywhere in a string might translate to “5th of Unodin in 937, at 7 o’clock (PM)”. Then Ylliakum times would be automatically readjusted after a server crash. (Translation would have to be done whenever text is stored in a book though, to avoid “auto-updating books”!)

I cannot enter umlauts in the Guild Message window or in chat.

This affects my German Ubuntu Dapper Drake machine, so I’ll leave this as a *nix bug until somebody confirms this for some other OS.

A selection of affected characters (those that happen to be available on my keyboard): äöüÄÖÜß²³€§µ°

²³€§µ° wouldn’t be a serious loss from an immersive experience point of view, but if that’s intended, the game should also filter @^$\~#|%&=

How to reproduce:

Boot client machine while disconnected from the network.
Start PS client.
Client lists the Fragnetics server as “Failed” (OK).
Reconnect client to network.
Client continues to list Fragnetics as “Failed”. It should detect that the server has become available.

Workaround:
Restart client.

Speculation:
I suspect this is because the client caches the response of the initial DNS query, which returns NXDOMAIN in the described situation.
An alternative cause might be that the networking code somehow already knows the IP address but gives up permanently due to a “no route to host” error message.

Suggestion:
Make the networking code generally re-request DNS information whenever the server becomes unreachable.
This would not only cover the situation described above, but would also remove one obstacle to running the PS server on a dyndns-managed host.