Whenever a lua script references a core struct obj, increment a counter
in the obj struct. Core code will not free the obj, if there are any
lua references pointing to it, just makes it free-floating.
When lua script ends, the lua gc will free the free-floating objects.
Also exposes u.inventory to lua.
Breaks save and bones compat.
Meat rings were causing increased hunger even though they don't do
anything. Not mentioned in the report, but cheap plastic imitation
amulets increased hunger too and they don't do anything either.
Trickier to fix, +0 rings of protection were excluded from hunger
on the grounds that +0 rings don't do anything, but since 3.6.0
+0 protection provides +1 to magic cancellation if protection isn't
coming from anywhere else. Two +0 rings of protection are trickier
still since only one of them provides the MC bonus.
Fixes#272
After the "make 'w' parallel with 'Q'" patch, wielding bare hands
was erroneously treating object id 0 as a split of zeroobj. That
isn't in inventory so seems 'lost'. Fixed by testing for nonzero.
There was another bug: you could wield a partial stack even if your
current weapon was cursed. Fixed by reordering if/else-if/end-if.
Change the composition of worm tooth from none-of-the-above to bone
and crysknife from mineral to bone, same as is used for unicorn horn.
I think the only significant difference will be that worm teeth used
up during polypiling will produce skeletons rather than flesh golems.
Fixes#268
Originally requested by one of the hardfought admins
Adjust all active window ports (tty, curses, win32, Qt, X11) to store
the itemflags that they receive with each item.
Also, make those active window ports understand the new
MENU_ITEMFLAGS_SKIPINVERT flag by skipping any menu items with that
setting during invert_all and invert_page operations.
Build testing and rudimentary functionality testing was carried out
on each of the window ports listed above.
The code was also modified on some non-active window ports (Qt3, gem,
gnome) but it was not tested for build or function there.
The desired functionality expressed was to be able to select a
single object category, and use the @ "invert all" function to
exclude that one and select all the others.
The "invert all" function's behavior of also including things
like "select all" and BUCX menu items made the feature unuseful
for that purpose.
Increase weight of giant spider from 100 to 200; leave nutrition at 100.
Increase weight of giant beetle from 10 to 200; increase nutrition from
10 to 50. Both are still size 'large'.
I've left giant ant with weight 10, nutrition 10, size 'tiny' so that
it doesn't become bigger than soldier and fire ants.
Fixes#267
Subtracting one dungeon depth value from another had the subtraction
backwards and that yielded a negative value where a positive one is
expected. If NH_RELEASE_STATUS were to be set to NH_STATUS_RELEASED
then this was at risk of crashing (if the bad subtraction yields -2,
rn2(diff+2) would divide by 0) since rn2()'s argument isn't validated
for released version.
fixes37.0 was confused, listing a couple of things that aren't bugs
in 3.6 as general fixes. I suspect that the DLB one was fixed before
being exposed via git, so shouldn't be there at all.
groundwork only - window port interface change
This changes the last parameter for add_menu() from a boolean
to an unsigned int, to allow additional itemflags in future
beyond just the "preselected" that the original boolean offered.
There shouldn't be any functionality changes with this groundwork-only
change, and if there are it is unintentional and should be reported.
If you're wielding a stack of N items, issuing the command to quiver
them asks whether you want to quiver N-1 of them (implicitly leaving
one wielded). If you answer no then you're asked whether to quiver
all of them. You could also give a count when picking the item to be
quivered and the stack would be split based on that.
However, if you have a stack of N items quivered, issuing the command
to wield them just did so, leaving the quiver empty. And picking an
item ignored any count, so even explicitly asking for 1 (out of N)
wielded the whole stack. Change 'w' to parallel 'Q'; if you try to
wield a quivered stack, you'll be asked whether to wield just 1 of
them. For no, ask whether to wield the whole stack. Or you can give
an explicit count when picking any stack in inventory to wield.
Both 'w' and 'Q' probably ought to handle the alternate/secondary
weapon similarly when it contains a stack. This doesn't address that.
Bite the bullet and add a special purpose boolean option to control
game behavior for random clairvoyance. When objects or monsters are
discovered, it normally issues "you sense your surroundings" and
performs a getpos() operation which allows the player to browse the
map by moving the cursor around and getting 'autodescribe' feedback.
But there have been complaints that once the hero has the Amulet
(which triggers random clairvoyance even though hero isn't flagged
as having that attribute) the message and pause-to-browse become too
intrusive.
This was initially combined with the 'timed clairvoyance' fix because
they both bump EDITLEVEL to invalidate existing save files, but their
details don't interact so I separated them.
When the hero has random clairvoyance, the code used
| (moves % 15) == 0 && rn2(2) != 0
(where 'moves' is actually the turn number) to decide when it would
kick in and show a portion of the map. If the hero was fast enough
to get an extra move when the turn value met the (moves % 15) == 0
condition then clairvoyance could happen twice (or more if poly'd)
on the same turn.
The changes (one new field, reordering a few others) in 'struct
context' invalidate existing 3.7.0-x save files.
Fixes#266
Build feedback filtered by a script which filters out -Dthis -Wthat:
gcc -g -I../include -I../lib/lua-/src -c ../win/share/tilemap.c
The second -I is obsolete or else its bogus value would have caused
build failure. When removing it, I noticed that there was still quite
a bit of obsolete yacc and lex stuff in there. Remove that too.
