While an object is being thrown, it isn't on any list. This means that
killing a shopkeeper with an unpaid object wouldn't be able to clear the
unpaid bit. By the time the object lands, the shopkeeper is gone, and then
it's too late. Added a new global to track a thrown object, set it and later
clear it in throwit(), also clear it as needed in dealloc_obj(), and check
it in setpaid(). It should be possible to use this global to avoid losing
thrown objects during hangup saves as well. But that can wait.
Building with an old version of gcc with various warnings enabled
generated a lot of noise. Most of it was due to not guarding string
literals with `const', but there were a couple of actual problems too.
Fix the reported problem of getting impossible warnings for unpaid
inventory items during final disclosure if you level teleport out of
the dungeon while holding shop goods. The fix makes you trigger a shop
robbery first; that required a minor end of game change to prevent the
shopkeeper from seeming to tag along (shk would take your possessions
as if you'd died on the shop level).
From a bug report. I haven't looked closely, but I think this bug is just
in the development code. If you dig in a shop with a wand while carrying
unpaid stuff, the shopkeeper would grab your pack, including the wand but
w/o clearing current_wand. It seems the right thing is to just not grab
the current_wand, since it's not in your pack while you're zapping it.
In 3.4.0, you lost the wand too.
Reported to the list. Player breaks a wand of digging causing the
shopkeeper to fall. This didn't clear unpaid items. For now, this can
only occur due to player action, so added a call to make_angry_shk to
handle this situation.
- Breaking wand of digging dug through rock which should be undiggable.
Checks assumed pits would never show up in solid rock.
- Breaking wand of digging near shop walls wouldn't anger the shopkeeper
Checks assumed pits would never show up in walls, also, added a special
case to pay_for_damage to handle the case where you're falling thru and
can't be asked to pay.
- Shop walls wouldn't be restored if there are pits in the way.
Checks assumed pits would never show up in walls.
- If there was a hole outside the shop, you could kick stuff out of the
door into the hole without shopkeeper noticing. Added the missing check.
Change the message for paying off consumed items to
"You paid for a xxxx at a cost of nnn gold pieces."
to distinguish it from objects that you still have in your inventory which is
"You bought a xxx for nnn gold pieces."
The worn bits, et al, were not all cleared when the shopkeeper grabbed your
pack. This currently affects only alternate (non-twoweap) and quivered
objects. This could lead to a panic or a crash.
[ Fix a bug reported in the newsgroup; I thought I sent this last
week, but it isn't in the current code so I must have forgotten. ]
When I split u_left_shop() into two routines I neglected to
propagate the early return condition from the second half to the
first. The result is that if you leave a shop with unpaid goods
but have enough credit there to cover the cost, the shopkeeper
will take that credit and be satisified, but the kops were still
getting summoned as if he had been robbed.
- when a shopkeeper leaves the shop to chase the player, and the player
enters the shop, bill_p is set to an unusual value. bill_p needs to be set
back to a valid value if the shopkeeper re-enters the shop.
- Also, the u.ushops state needs to be updated when a shop becomes tended
again if the player is in the shop.
- introduce a new after_shk_move function to handle this
There was at least one more special case aside from throwing
(jetisoning items to reduce weight after falling in water) which
have needed the same extra code. This is a more general fix.
- new cxname() to simplify doing the right thing in increasingly common cases
- use for bullwhip snagging
- in shopkeeper offer code
- in a couple other existing places rather than duplicating CORPSE checks
- use singular(...) in "swings" cases, since only one can hit. Singular uses
corpse_xname automatically when appropriate
includes container contents, not just the cost of the
container itself (a prices in inventory phenomenon).
Along the way I discovered a peculiarity -
contained_cost() was adding up the cost of everything in
a container, even if you had stashed items in it that were your
own and not marked unpaid it seems.
I added a flag to force the code to only add objects
that were marked "unpaid" so I could use it in this new
instance, but I didn't change any of
the existing usages (I left the flag at FALSE which leaves
the consideration of the unpaid status alone just as
before).
Some of this is correction of some messages that were
wrong prior to this when dealing with selling of objects
inside a container when only part of the contents was unpaid.
The earlier patch made sure that bribe() didn't pass an
invalid value to money2mon(). This one changes money2mon() so
that if some other code else does so, reporting the impossible
situation won't be followed by a splitobj panic. Most of this
patch is reformatting though.
over the place.
Often they would use
"%ld zorkmid%s", amt, plur(amt)
but not consistently, so some of the hard-coded usage
could result in "1 zorkmids"
This adds the function
currency(long)
to return the name of the currency, either plural
or singular depending on the argument passed to it.
That eliminates the need for the extra %s in the
format string and the use of the plur() macro.
If you get interrupted while reading a spellbook and then
the book gets destroyed or you change levels, the object pointer
remembered for the book will be invalid and could accidentally
match one subsequently allocated to some other book. That would
result in "you continue your efforts to memorize the spell" when
starting to read that other book; it would also end up bypassing
the reading difficulty check and reuse the old book's delay counter.
I don't remember who reported this. It was quite some time
ago and I have an abandoned patch dated last March from when I
first started to fix it.
Files patched:
include/extern.h
src/save.c, shk.c, spell.c
auto_credit flag; that's what sell_response is for).
follow-up comment from Pat:
This introduces behavior that I consider to be buggy.
You need something separate from sell_response, otherwise if
you answer (a)ll when the shopkeeper still has money you end
up selling everything for credit without being asked when he
runs out of cash in the middle of the transation. Avoiding
that is the reason for the old behavior in the first place.
