The g? structs had a mix of variables that were written to
the savefile, and those that were not.
For better clarity and to distinguish those that end up in
the savefile, relocate some g? variables that get written
directly to the savefile into different structs.
This updates EDITLEVEL, although technically it probably
didn't need to, since savefile contents are not changing.
Details:
gb.bases -> svb.bases
gb.bbubbles -> svb.bbubbles
gb.branches -> svb.branches
gc.context -> svc.context
gd.disco -> svd.disco
gd.dndest -> svd.dndest
gd.doors -> svd.doors
gd.doors_alloc -> svd.doors_alloc
gd.dungeon_topology -> svd.dungeon_topology
gd.dungeons -> svd.dungeons
ge.exclusion_zones -> sve.exclusion_zones
gh.hackpid -> svh.hackpid
gi.inv_pos -> svi.inv_pos
gk.killer -> svk.killer
gl.lastseentyp -> svl.lastseentyp
gl.level -> svl.level
gl.level_info -> svl.level_info
gm.mapseenchn -> svm.mapseenchn
gm.moves -> svm.moves
gm.mvitals -> svm.mvitals
gn.n_dgns -> svn.n_dgns
gn.n_regions -> svn.n_regions
gn.nroom -> svn.nroom
go.oracle_cnt -> svo.oracle_cnt
gp.pl_character -> svp.pl_character
gp.pl_fruit -> svp.pl_fruit
gp.plname -> svp.plname
gp.program_state -> svp.program_state
gq.quest_status -> svq.quest_status
gr.rooms -> svr.rooms
gs.sp_levchn -> svs.sp_levchn
gs.spl_book -> svs.spl_book
gt.timer_id -> svt.timer_id
gt.tune -> svt.tune
gu.updest -> svu.updest
gx.xmax -> svx.xmax
gx.xmin -> svx.xmin
gy.ymax -> svy.ymax
gy.ymin -> svy.ymin
Related note:
There are some pointer variables that are heads of chains that were not
moved from 'g?' to 'sv?', because they are not actually written to the
savefile directly, but the objects/monst/trap/lightsource/timer in the
chains they point to are. That can be changed, if desired.
Examples: gi.invent, gm.migrating_objs, gb.billobjs, gm.migrating_mons,
gf.ftrap, gl.light_base, gt.timer_base
- add nhl_pcall_handle() to wrap all nhl_pcall calls that didn't check
return value and either panic() or impossible()
- add --loglua (unix only) to dump Lua memory and steps info to livelog
- remove old logging
- set memory and step limits on all Lua VMs
selection_getbounds() has a check and early return.
Initialization will ensure a known state if that early return
were ever taken.
This is an alternative approach to pr #1163.
Something that's reasonable to expect to see in Lua files is something
like:
local sel4 = sel1 - sel2 - sel3
or more generally, producing a selection from subtraction that will then
be used in subsequent selection math.
I discovered this wasn't actually working correctly, and that it also
applied to the xor operation. The reason behind this is that
l_selection_sub and l_selection_xor create a new selection from nothing,
which by default has "lower" bounds of COLNO, ROWNO and "upper" bounds
of 0,0. Iterating across the intersecting rectangle of both selections
does not reliably set the bounds of the resulting selection properly,
since the first selection_setpoint with a value of 0 will cause the
selection's bounds_dirty flag to be set, at which point they will cease
to change as more points are added.
Then this selection with its incorrect boundaries is pushed back onto
the Lua stack, and becomes the first operand of the next subtraction
(i.e. selr in the first l_selection_sub becomes sela in the second
l_selection_sub). Depending on how broken the bounding box is, results
may vary, but if the bounding box is still (COLNO,ROWNO,0,0), the
resulting selection will have no points selected at all.
This fixes this problem by forcibly recalculating the bounds of the
result selection, so any subsequent operations on it will be valid.
selection.gradient has some pretty unintuitive behavior, in that it
selects points that are NOT close to the defined center. I've used
gradient selections several times and so far all of them have had to be
negated, because I wanted to select points close to the center with a
decreasing probability further out.
This implements that behavior, and also fixes a bug in which the x,y
coordinates of the gradient center(s) were not converted properly when
used within a des.room or des.map. Also updated the lua documentation
for gradient.
I removed the "limited" argument, as it was previously used to control
whether the rest of the map outside the max given distance would be
included in the selection; now that the area beyond maxdist is naturally
never in the selection, it doesn't have much use. (And I can't think of
a reasonable use case for the inverse: wanting to select points close to
the center, with decreasing chance towards maxdist, but then select the
entire map beyond maxdist.)
Currently this does not affect any special levels or themed rooms
because none of them use selection.gradient.
The lava river will now draw another river, until a certain
amount of map locations have been turned into lava, so you don't
get a teensy "river" made out of 2 lava pools.
Add a lua selection method to count the number of locations
in the selection.
Instead of just plain old boring mazes, spice up Gehennom by
occasionally adding lava, iron bars, or even mines-style levels
(with lava, of course).
Of the fixed Gehennom levels, only Asmodeus' lair has been changed
to add some random lava pools.
Also some lua fixes and changes:
- Fixed a selection negation bounding box being wrong.
- Fixed a selection negated and ORed returning wrong results.
- des.map now returns a selection of the map grids it touched.
- When using des.map contents-function the commands following the
map are not relative to it.
The consolidation of global variables from scattered source
files into decl.c and declared in decl.h was begun in 3.7.0.
Their placement in common files was done for centralized
initialization and potential re-initialization during a
"play again" scenario.
It wasn't really necessary for all of them to be housed in a
single huge structure to meet the "play again" requirement,
and the single huge structure has been a little unwieldy when
it comes to maintenance.
Following this commit, instead of one single extremely large structure
named 'g' to house all of the relocated global variables, they
are distributed into several ga through gz.
To make things easy for the developer, each variable is placed
into the struct corresponding to the starting letter of the variable.
That way, no lookup is required in order to know which struct houses
a particular variable, it is a simple match to the starting letter
for all the centralized global variables.
A global variable named 'amulets', would be found in ga.
ga.amulets
^ ^
A global varable named 'move', would be found in gm.
gm.moves
^ ^
A global variable named 'val_for_n_or_more' would be found in gv.
gv.val_for_n_or_more
^ ^
A global variable named 'youmonst' would be found in gy.
gy.youmonst
^ ^
Although gcc specifies support for declaring a function as
noreturn after the function name and parameters, other compilers
do so via an attribute at the start of the declaration. Add some
macro support for the attribute-at-the-beginning method:
o MS Visual Studio compiler
o Upcoming C23 standard (untested at this point)
Previously, the tetris-shaped rooms were always either
normal rooms, or turned into shops or other special rooms
in NetHack core. Now, the themed room lua code first picks
the themed room (which can be a themed or shaped), and some
of those will then pick a random filling (eg. ice floor,
traps, corpses, 3 altars).
Adds a new lua binding to create a selection picking locations
in current room.
The content-function in special level regions now get passed
the room data as a parameter.
This is a large iteration on a previous implementation of making
nh.getmap() parse its coordinates as relative to the last defined map or
room rather than absolute to the entire level. Now, everything in the
nh.* and obj.* functions interprets coords as relative rather than
absolute. (By default; if no map or room has been defined, or if the lua
code is executing after level creation is done, they will interpret the
coordinates as absolute).
The general motivation is basically the same - routines that use
absolute coordinates are difficult to use in level creation routines,
because then the designer has to remember to convert the relative
coordinate to an absolute one (and that was impossible before
nh.abscoord was added, particularly in themed rooms). And once
nh.getmap() takes relative coordinates, it would be very strange to have
all the other functions (setting timers, burying objects, etc) remain
with absolute ones.
In a couple places, code is changed to account for coordinates that are
relative to a *room* (which uses g.coder->croom->[lx,ly] as an offset,
instead of relative to a *map*, which uses [xstart,ystart].
Specifically, selection.iterate did not account for this, and without
this the ice themed room timer was not being started in the proper
place.
All tests are updated to respect the new behavior. Most of the modified
functions are not actually used anywhere in level files; the one
exception is starting a timer in a themed room, and that has been
adjusted.
Documentation updated as well to clarify when various things are tossing
around relative and absolute coordinates, both in comments and in
lua.adoc.
- Add bounds, so that we don't process any locations outside
as those locations are known to be unset
- The bounds are only recalculated if needed
- Replace instances of selection_not where we actually want
a new selection with all locations set
One of the drivers of this change was that screen coordinates require a
type that can hold values greater than 127. Parameters to the window
port routines require a large type in order to be able to have values
a fair bit larger than COLNO and ROWNO passed to them, particularly for
their use to the right of the map window.
This splits the uses of xchar into 3 different situations, and adjusts
their type and size:
xchar
|
-----------------------
| | |
coordxy xint16 xint8
coordxy: Actual x or y coordinates for various things (moved to 16-bits).
xint16: Same data size as coordxy, but for non-coordinate use (16-bits).
xint8: There are only a few use cases initially, where it was very
plain to see that the variable could remain as 8-bits, rather
than be bumped to 16-bits. There are probably more such cases
that could be changed after additional review.
Note: This first changed all xchar variables to coordxy. Some were
reviewed and got changed to xint16 or xint8 when it became apparent that
their usage was not for coordinates.
This increments EDITLEVEL in patchlevel.h
This is enough to prevent abuse by denying access to functions and
denial of service (RAM and instruction step limits), but not enough
to allow restricted use of things that require finer control (e.g.
filesystem access).
If something goes wrong, the whole thing can be turned off, for
now, in config.h (see NHL_SANDBOX).
None of the current functionality requires changes to build systems;
some of the possible future functionality may require some #defines
- TBD. There is lots of dead code (#ifdef notyet) for bits of that
additional functionality; we can rip it out if we don't want those
additions or we can complete (parts of) it depending on our needs.
All current uses of Lua are connected to sandboxes and guarded with
nhl_pcall (sandbox and lua_pcall wrapper); options and limits can
be set at the callsites in the passed nhl_sandbox_info. Some of
the error handling may be wrong - panic() vs. impossible() vs
silence.
Memory and instruction step limits should be tuned prior to release;
there's no point tuning them now.
Make selection rndcoord return a table with x and y keys.
Allow (most) coordinate parameters accept such a table.
Fix selection and des lua tests broken by the above changes and
an earlier change, because selections tried to set terrain
at column 0, and it now causes a complaint.
Noticed that when I set a selection to grow in a random direction, it
instead grew in all directions, which is not what I wanted. Turns out
the -1 random dir ended up being passed straight to the code which
checks bitmasks, without any form of randomizing among directions.
So this adds code to do that, and defines W_RANDOM as -1 rather than
using a magic number. In the process I also noticed that specifying
"random" as the wall for a door in a room made it rerandomize the
direction every iteration of its loop, essentially rolling two rn2(4)s
and only proceeding if they matched. That was pointless so I cleaned it
up a bit.
Also added safety checks in the form of an impossible for des.corridor()
being called with "random" as either walldir, because this is not
implemented currently.
And lastly, I noticed that create_secret_door was entirely unused
(secret door creation is handled in create_door), so I deleted it.
The only behavior change caused by this is that the Valkyrie quest lava
pools will be a little smaller, which is the only place grow is
currently used. If it's desired to keep them the same, that should be
changed to "all".
Selection difference is something I have found myself wanting a lot when
working on levels, and have had to defer to a clunkier xor-then-and
approach. This commit implements the TODO-ed addition and subtraction
operators on two sets.
I don't see how the addition operator would be any different from
logical or, so it just calls l_selection_or rather than implement a new
function.
../src/nhlsel.c(467) : warning C4701: potentially uninitialized local variable 'x1' used
../src/nhlsel.c(591) : warning C4701: potentially uninitialized local variable 'x' used
../src/nhlsel.c(467) : warning C4701: potentially uninitialized local variable 'y1' used
../src/nhlsel.c(591) : warning C4701: potentially uninitialized local variable 'y' used
../src/nhlsel.c(469) : warning C4701: potentially uninitialized local variable 'x2' used
../src/nhlsel.c(469) : warning C4701: potentially uninitialized local variable 'y2' used
If we ever want huge maps with COLNO or ROWNO larger than signed char,
this will at least allow the game to compile and start when typedef'ing
xchar to int. Trying to use huge maps exposes more bugs.
Uncomments and makes available selection.gradient in Lua. (The backend C
code for this still existed, it just wasn't used.)
The only valid way to specify a gradient is with a table. I considered
adding non-table versions, but decided that there are too many
independent variables that can be optional. A non-table version, without
named parameters, would be confusing to read, especially since most of
the arguments are ints.
Also adds an impossible in the (possibly unreachable) case that
selection_do_gradient gets called with a bad gradient type.
There was a rare selection bug where selection was freed by the gc
but it was still in use. Don't remove the selections from the stack
while we're handling them.
A recent change has resulted in complaints about 'sel' not being
used in
struct selectionvar *sel = l_selection_check(L, 1);
struct selectionvar *ret;
int p = (int) luaL_checkinteger(L, 2);
lua_pop(L, 1);
(void) l_selection_clone(L);
ret = l_selection_check(L, 1);
selection_filter_percent(ret, p);
This preserves the semantics of that code while getting rid of the
warning, but I have no idea whether it makes any sense.
