Refine the fix #K3888 by moving multiple instances of the check for
cancel_don() from destroy_arm() to one instance in wornarm_destroyed().
Should be no change in behavior.
Report was for spell-casting monster using the destroy armor spell on
hero's levitation boots while hero was floating over lava. The boots
became unworn but still in inventory, hero dropped into lava, the
boots happened to be an inventory item which got burned up, then the
call stack unwound back to the destroy armor routine which tried to
finish by deleting them but they were already gone by then. Could
also happen for black dragon breath, hero reading scroll of destroy
armor, or overenchanting the boots with scroll of enchant armor, so
not so unlikely that nobody would be expected to notice.
Initially I couldn't reproduce the object lost panic. It only happens
if the memory used by the boots is cleared or clobbered during first
time it's freed, otherwise second free doesn't notice any problem.
The 'wornarm_destroyed()' portion of this commit is sufficient to fix
this. The other bits are things I tried before figuring out how to
reproduce it, plus zeroing out any object passed to dealloc_obj().
Defining des.object({ id="large box", locked=false })
was the same as random locked state. Make it actually mean unlocked,
and not defining locked at all means random.
To get the Magic Mirror of Merlin to speak, you could apply it in
any direction (or wield it). To get the Master Key of Thievery to
speak, you had to apply it toward an adjacent doorway or down while
on a container (or wield it). Make the key behave like the mirror.
When a genetic engineer polymorphs someone it normally teleports away.
Also set mspec_used so that it can't polymorph someone [else] on its
next turn, it case of a no-teleport level or it happens to randomly
land adjacent to the target.
Previously the fuzzer usually stayed in the Dungeons of Doom,
only very rarely going to other dungeon branches. Now, it will
randomly choose a dungeon branch and a level in that branch
to level teleport to.
Special level creation could make levels with boulders on top
of lava or water; this was caused by mazewalk populating the maze
before the rest of the level was created.
Add a post-level-creation map cleanup routine, where boulders
and traps on liquid terrain are removed.
If asked to create an Angel when MM_EMIN isn't specified, makemon()
will either create an ordinary monster of type Angel without any
emin extension or add emin and create a minion affiliated with the
god for randomly chosen alignment. Figurines didn't specify emin
so could result in either type, but wishing for them offered no
opportunity to specify alignment for the second type.
Change figurine activation to force the first type: an ordinary
monster of type Angel, not affiliated with any particular deity.
Add "again" to the hit message when an attack sequence has
consecutive attacks of same type and they hit. A {bite, claw, claw}
sequence won't give that for first claw attack regardless of whether
the bite hits but will give it for the second claw attack when both
of the claw attacks hit.
The message sequence
|The fire ant bites! The fire ant bites! You're on fire!
or |The fire ant bites! The fire ant bites! You avoid harm.
when the first bite was for physical damage and the second was
for fire damage seemed a little confusing.
This changes that to be
|The fire ant bites! The fire ant bites again! You're on fire!
or |The fire ant bites! The fire ant bites again! You avoid harm.
It still isn't crystal clear that both bites are from a single
attack and that the second bite is for different type of damage
but I think it's an improvement.
Many damage handling routines were calling mhitm_mgc_atk_negated()
in advance so that the result could be used for u-vs-m and m-vs-u,
and m-vs-m variations of an attack. But the monster versus you case
called hitmsg() to deliver a "<mon> <bites, claws, &c> you" message
after that. When the negation checking routine recently began
issuing messages for some types of damage, they would be delivered
before that hit message when your armor/jewelry negated its damage.
|You avoid harm. [when MC is about to block the shock]
|The grid bug bites. [bite for electric damage]
or
|The fire ant bites. [for physical damage]
|You avoid harm. [when MC is about to block the fire]
|The fire ant bites. [second bite for fire damage]
This fixes the sequencing issue at cost of making the code become
even more complicated. It will probably require further refinement.
Be more consistent with the engulf attack feedback by creatures who
fold themselves around the victim (trapper, lurker above) rather than
swallow or directly engulf.
Replace an instance of a non-literal format string and the warnings
manipulation it needed with a literal one.
Adjust the message given when an attack knocks its target back. Say
|<defender> is knocked backward by <attacker>
if target will actually move or
|<defender> is knocked back by <attacker>
if there's something preventing the move. Most players probably won't
even notice the difference. (Possibly "rocked back" would be better
when not changing location but this hasn't gone that far.)
Also make the knock back distance be 1 step 2/3 of the time or 2 steps
1/3 instead of 50:50 chance for 1 or 2 steps.
I happened to notice when looking at a ttyrec that farlooking
with automatic description would blink the cursor on hero for
one frame, and then back on the farlooked map location.
Using #retravel when no destination was set up just silenty finished,
effectively a no-op. Could happen for the first travel attempt on
the current dungeon level, or after most recent #travel successfully
reached and cleared its destination.
Using #retravel when already at previously set but unreached--via
travel--destination, or explicitly picking your own spot as travel
destination using #travel, actually traveled from your current spot
to the same spot. Usually nothing interesting occurred and you got
no feedback. However, if you happened to be stuck in a trap at that
spot you tried to walk out of it and got trap feedback about that.
(Then probably right back into the trap if you succeeded in escaping
it. I didn't try to test that.)
Extend "killed by the touch of death inflicted by <monster>" to buzz().
"Killed by a bolt of cold" becomes "killed by a bolt of cold zapped by
<monster>" or "killed by a blast of cold" becomes "killed by a blast
of cold exhaled by <monster>" and so forth.
More work than expected; the zap code isn't passed enough context.
BZ_M_WAND() was producing the wrong value for wands zapped by monsters.
Doors weren't getting added to the correct subrooms in certain cases.
Also fix one of the themerooms, because doors have to be added
after subrooms; there was a possibility of no door to the subroom(s)
in that themeroom, because the subrooms overwrote the doors in
the parent room.
Test case for the subroom doors:
Large room, with a medium subroom, with a tiny subroom inside that.
The doors go from outermost room <-> tiny innermost room <-> middle room.
des.room({ type = "ordinary", x = 1, y = 1, w = 10, h = 10,
contents = function()
des.room({ type = "ordinary", w = 6, h = 6, x = 2, y = 2,
contents = function()
des.room({ type = "ordinary", w = 2, h = 2, x = 0, y = 0,
contents = function()
des.door({ state="random", wall="south", pos = 1 });
end
});
des.door({ state="random", wall="north", pos = 1 });
end
});
end
});
Before this fix:
ROOM: ndoors:1, subrooms:1
SUBROOM: ndoors:1, subrooms:1
SUBROOM: ndoors:1, subrooms:0
after this fix:
ROOM: ndoors:1, subrooms:1
SUBROOM: ndoors:1, subrooms:1
SUBROOM: ndoors:2, subrooms:0
This is fairly ridiculous but it prevents the bogus complaints
when compiling eat.c with 'gcc -fanalyzer' about some fields in
gc.context.victual being used without having been initialized.
There's bound to be a better way to handle this and I'm curious
whether it will work with the 'onefile' testing.
Issue reported by vultur-cadens: cause of death reason for touch
of death and death due to loss of strength only showed the cause,
not the monster spellcaster who was responsible.
This changes
|Killed by a touch of death.
to
|Killed by the touch of death inflicted by the Wizard of Yendor.
and
|Killed by terminal fraility.
to
|Killed by strength loss inflicted by a chameleon imitating an arch-lich.
(The 'imitating' part doesn't fit on the tombstone but will be present
in logfile/xlogfile.)
Noticed while implemented this: touch of death was modifying u.uhpmax
and basing death vs damage on that even when hero was polymorphed.
It now rehumanizes the hero in that situation.
Closes#994
1. Add "engraved room floor" pchar sym (S_engroom). The symbol that
displays at the engraved part of a room (not a corridor though).
The default symbol is '`' which is currently never shown if people
have defined the boulder symbol to '0' and statues are displayed as
monster symbols. It is bright blue.
Add some stylized variations of the S_engroom symset to some of
the symsets.
2. Add "engraved corridor" pchar sym (S_engrcorr). The symbol that
displays at the engraved part of a corridor. The default symbol is
'#', and it matches the symbol for corridor from for whatever the
current symset uses. It is bright blue to match the color of the
S_engroom symbol. Using the normal corridor symbol for display
preserves the lines of the corridor so is not as visually-disruptive
as a smaller symbol would be. Explicit entries that match the S_corr
symbol have been added to the symset file.
Magic mapping and clairvoyance impacts yet to be determined.
The Guidebook updates will come later.
