Changes most of the special keys used in the main input loop
into extended commands:
- movement keys are now bound to extended commands, eg.
#movewest and so on.
- m-prefix is now #reqmenu extended command, still bound to
the 'm' key.
- run, rush, and fight are now extended commands, still bound
to the same keys as previously.
- nopickup and runnopickup keys are removed.
Nopickup was using 'm' key, the same as the m-prefix, so
allow #reqmenu to modify movement commands to disable pickup.
- multiple prefix commands are allowed. This lets user to
use #reqmenu, followed by #run, followed by movement to simulate
runnopickup behaviour. (If necessary, adding runnopickup back
as an extended command would be easy)
Instead of returning ECMD_OK, the commands now return ECMD_CANCEL
when user declined to pick a direction or an object to act on.
Note that this can be ORed with ECMD_TIME, if the command still
took a turn.
For now this has no gameplay meaning.
Always give a message when creating a detected monster
during gameplay (as opposed to during level creation).
To prevent the message, use the MM_NOMSG flag for makemon.
Most places already handled their own messaging, but there
were some, such as bag of tricks, create monster magic
and random monsters created during gameplay that didn't.
Instead of returning 0 or 1, we'll now use ECMD_OK or ECMD_TURN.
These have the same meaning as the hardcoded numbers; ECMD_TURN
means the command uses a turn.
In future, could add eg. a flag denoting "user cancelled command"
or "command failed", and should clear eg. the cmdq.
Mostly this was simply replacing return values with the defines
in the extended commands, so hopefully I didn't break anything.
Indent all labels one space. Having uniform placement makes spotting
them much easier. (Having no indent at all would impact the change
bars of 'git diff'. Those display the last unindented line--which
doesn't start with punctuation--occuring before each band of changes,
so usually the name of the function being changed now that we no
longer have unindented K&R-style function argument declarations.)
While in there, shorten or split various wide lines and replace a few
tabs with spaces.
Suppress any time spent in a sub-shell or in the background when
accumulating total elapsed play time.
This won't help for leaving the game idle instead of saving and
restoring. That's a can of worms I'd prefer to leave sealed.
(including resuming travel after being interrupted)
Attempting to travel from an 'interesting' spot that would normally
stop travel mid-attempt (next to a door, next to a monster, etc) would
fail to get off the ground, because the 'interesting thing' would halt
travel before taking the first step.
Prior to 433f0cc, initiating travel would immediately trigger a call to
domove in rhack; as part of refactoring travel, the order of events in
rhack was changed so that this didn't happen immediately as before. Add
domove to the end of dotravel_target to produce a similar effect to the
previous arrangement.
domove also zeroes out g.domove_attempting, which I think is the reason
c0c617c seemed to mitigate some of the issues associated with this
problem (e.g. travel target selection seemingly not registering right
away).
Fixes#559
Put the flag that parse() uses to tell readchar() that the next
keystroke is a command into the program_state structure instead of
having it be a static variable in cmd.c. Conceivably an interface
could make use of it, and even if none do, it is program state....
More #558
Setting the 'altmeta' option affects how ESC is handled. Normally it
is for terminals/emulators which transmit two character sequence ESC c
when the user types Meta+c, but it can also be used to construct meta
characters 'manually' by typing ESC c. Unfortunately setting this
option has a side-effect of requiring a second character after ESC
before our readchar() will return, so it is only honored when parse()
wants the next command, not for general input.
When readchar() was recently split into two parts, the use of static
variable 'alt_esc' by parse() and readchar() was removed. That
resulted in requiring the user to type a second character whenever
ESC was typed, instead of just when it was transmitted or typed as
the prefix of a command that uses a meta-character keystroke.
Instead of just putting 'alt_esc' back, do things differently by
adding static 'getting_cmd' flag instead.
Fixes#558
Add des.finalize_level() used for testing in conjunction with
des.reset_level().
Add nhc.DLB to return 0 or 1 if DLB was defined at compile-time.
Change the test_lev.lua to give more informative messages instead of
just lua error when required file doesn't exist.
Add bigrm-11 to the level tests.
The for loop which iterates through the list of movement keys in
movecmd(cmd.c) was updated in 5abf948116 to count down to 0 instead of
up to the end of the list. This commit inadvertently introduced an
off-by-one error which started the loop one past the actual end of the
array. On my system this made 'H' stop working as the 'run West' key.
Retravel travels to the previously selected destination.
Also changes the travel-via-mouse to execute the extended command
instead of faking a special key.
Put the rush and run movement keys into g.Cmd instead of bit twiddling
the normal walk keys in multiple places to get the run and rush keys.
Allow meta keys in getpos. Use the normal running keys to fast-move
in getpos, instead of explicit HJKL - I polled couple places online,
and number_pad users did not use the HJKL keys in getpos.
Make meta keys work even after a prefix key.
Allows the fire-command to autowield a launcher; it will now
do either swapweapon or wield an appropriate launcher, if you
have ammo quivered.
This assistance can be turned off with the fireassist boolean option.
Adds a rudimentary command queue, which allows the code to add keys
or extended commands into the queue, and they're executed as if
the user did them. Time passes normally when doing the queue,
and the queue will get cleared if hero is interrupted.
Have curses call the core get_count() routine instead rolling its
own so that backspace and delete are supported. That part was
trivial to accomplish. Unfortunately it brought the disappearing
menu phenomenon back so it became more complicated overall.
Quit is not a commmand you usually need very often, and generally
don't want to use by accident.
Apparently, on Spanish keyboard layout, n-with-tilde is interpreted
by the Windows NetHack as M-q, and the key is next to l. Loot also
asks for confirmation, just like quit.
Prevent stuff like this by not binding the quit command to any key.
When the fuzzer is running, don't allow a randomly generated M-X
keystroke to switch from debug mode to explore mode.
[Unintended side-effect of the combination of two earlier changes:
assigning M-X as default key for #exploremode and allowing someone
in debug mode to voluntarily downgrade to explore mode (which should
never impact normal play and makes some types of testing simpler).]
Fixes#512
The #twoweapon command was flagged as autocomplete back when using
an extended command was the only way to execute it. Take that off
since simple 'X' suffices. Do the same for wizard mode commands
that can be invoked with control characters. Probably ought to do
the same for #overview too but this change doesn't.
I started to add the autocomplete flag for #exploremode but that
would require an extra letter for #enhance so I decided not to.
There are some wizard mode commands that can't be executed under
X11 because they aren't flagged to autocomplete and its extended
command selection widget only offers autocomplete commands as
choices. I haven't attempted to change that.
Always require paranoid confirmation for #panic rather than just
when it has been enabled for #quit.
Report from roughly two and half years ago was about "<monster>
opens the door" without displaying <monster>.
Monster movement first decides whether a monster can pass closed
door. If so, the monster is placed at the door spot, a message
is given about that movement (unlock, open, smash down, &c), and
finally the map is updated.
Changing the sequence to update the map before issuing the door
message was not sufficient to fix this. In the corridor plus
closed door plus lit room map fragment shown here, when 'O' moved
to '+', you would see it there if the hero is at '1' or '2', but
not if at '3', '4', or '5'; open door was shown instead. But the
message described 'O' accurately rather than as "it" for all those
hero locations.
: -----
: #O+1...
: |2...
: |3...
: |4...
: |5...
: -----
For 3,4,5 the #vision command shows the closed door as 3 before
the O move, but blank (0) after. In other words, the closed door
is within line of sight but once opened, the doorway spot isn't.
It makes sense that the closed door behaves like a wall but I'm
not sure whether the behavior for an open door's breach does too.
I had an awful workaround that successfully displayed the monster,
but it wouldn't show the same thing if the door was already open,
so I've changed the situation to yield "You see a door open."
The pull request that fixed a couple of instances where it was
possible to have multiple entries for gold in inventory indirectly
pointed out that the error checking was clumsy. If you executed
the #adjust command while having two '$' items in inventory, you
were told twice that you had multiple stacks of gold in inventory.
Change how that's handled so that the warning appears at most once
for any given #adjust command. Also avoids having #adjust's use
of getobj() re-scan entire invent for every item in invent.
Also, if player did manage to get two or more '$' entries, #adjust
would allow moving any but the last to a letter entry. Once in a
letter, further #adjust with count specified could split the letter
gold entries into even more gold entries. Now, if the player picks
gold as the #adjust 'from' item (which is only possible when there
are wrong letter gold entries or multiple ones or both) then #adjust
will now force 'to' slot to be '$' (without asking player to pick).
Lastly, the inventory check for multiple and/or wrong slot gold is
now performed by wizard mode sanity_check() in addition to #adjust.
Add new '|' command, aka #perminv, which allows the player to
send menu scrolling keystrokes to the persistent inventory window.
Implemented for X11, where its usefulness is limited, and for
curses, where it is more needed and also more fully functional.
The interface can either prompt for one keystroke, act upon it,
and return to normal play, or it can loop for multiple keystrokes
until player types <return> or <escape>. X11 does the former if
the 'slow' application resource is False so that prompting uses
popups, and the latter when 'slow' is True where prompting is in
a fixed spot and doesn't end up causing the persistent inventory
window to be stacked behind the map window. curses always does
the loop-until-done approach. It also accepts up and down arrow
keys to scroll one line at a time.
Also adds two new menu scrolling commands, menu_shift_right (key
'}' by default) and menu_shift_left ('{') if wincap2 flags contain
WC2_MENU_SHIFT. Shifting allows different substrings of too-long
lines to be seen.
For X11, neither works because their handling requires a horizontal
scrollbar and for some reason that escapes me our menus don't have
one of those. If they did, shifts could work for all menus but a
shifted window would hide the selection letters. So shifting would
be most usefully done as: pan right, read more of any long lines,
immediately pan back to the left.
For curses, they only apply to the persistent inventory window.
Shift right redraws it with class headers and inventory letters
shown normally but the item descriptions omit their leftmost
portion, showing more text towards the end. Shift left reverses
that and does nothing if the beginning is already in view. Forward
and backward scrolling while shifted leave the shift in place.
OPTIONS=menu_previous_page:\mv
BINDINGS=M-v:menu_previous_page
both worked, but
OPTIONS=menu_previous_page:M-v
BINDINGS=\mv:menu_previous_page
both failed. Make all four variations work. Tiny change made large
by the need to move some things around.
The option definition for menu_first_page had a couple of its flag
bits swapped. I didn't try to figure out whether that had any impact.
"Demote" wizmgender from an obscure wizard mode extended command
to an obscure wizard mode boolean option. Behaves the same except
that no message is given when the value gets toggled.
Whitelist all the verified existing triggers:
makedefs.c: In function ‘name_file’
attrib.c: one compiler balks at a ? b : c for fmtstring
cmd.c: In function ‘extcmd_via_menu’
cmd.c: In function ‘wiz_levltyp_legend’
do.c: In function ‘goto_level’
do_name.c: In function ‘coord_desc’
dungeon.c: In function ‘overview_stats’
eat.c: one compiler balks at a ? b : c for fmtstring
end.c: one compiler balks at a ? b : c for fmtstring
engrave.c: In function ‘engr_stats’
hack:c one compiler balks at a ? b : c for fmtstring
hacklib.c: one compiler balks at a ? b : c for fmtstring
insight.c: one compiler balks at a ? b : c for fmtstring
invent.c: In function ‘let_to_name’
light.c: In function ‘light_stats’
mhitm.c: In function ‘missmm’
options.c: In function ‘handler_symset’
options.c: In function ‘basic_menu_colors’
options.c: In function ‘optfn_o_autopickup_exceptions’
options.c: In function ‘optfn_o_menu_colors’
options.c: In function ‘optfn_o_message_types’
options.c: In function ‘optfn_o_status_cond’
options.c: In function ‘optfn_o_status_hilites’
options.c: In function ‘doset’
options.c: In function ‘doset_add_menu’
options.c: In function ‘show_menu_controls’
options.c: In function ‘handle_add_list_remove’
pager.c: In function ‘do_supplemental_info’
pager.c: In function ‘dohelp’
region.c: In function ‘region_stats’
rumors.c: sscanf usage
sounds.c: In function ‘domonnoise’
spell.c: In function ‘dospellmenu’
timeout.c: In function ‘timer_stats’
topten.c: In function ‘outentry’, fscanf, sscanf, fprintf usage
windows.c: In function ‘genl_status_update’
zap.c: one compiler balks at a ? b : c for fmtstring
win/curses/cursstat.c: In function ‘curses_status_update’
win/tty/wintty.c: In function ‘tty_status_update’
win/win32/mswproc.c: In function ‘mswin_status_update’
It turns out that macOS barked when a POP was issued without
a prior PUSH, so since the DISABLE_WARNING_CONDEXPR_IS_CONSTANT
expanded to an empty macro on that platform.
Include a corresponding
RESTORE_WARNING_CONDEXPR_IS_CONSTANT macro for use with that
particular warning.
Microsoft and other non-GNU compilers don't recognize gcc tricks
like /*NOTREACHED*/ to suppress individual warnings. clang recognizes most
of them because it tries to be gcc-compatible. Because of that, a lot of
potentially useful warnings have had to be completely suppressed in the
past in all source files when using the non-gcc compatible compilers.
Now that the code is C99, take advantage of a way to suppress warnings for
individual functions, a big step up from suppressing the warnings
altogether.
Unfortunately, it does require a bit of ugliness caused by the
insertion of some macros in a few spots, but I'm not aware of
a cleaner alternative that still allows warnings to be enabled
in general, while suppressing a warning for known white-listed
instances.
Prior to the warning-tiggering function, place whichever one of
the following is needed to suppress the warning being encountered:
DISABLE_WARNING_UNREACHABLE_CODE
DISABLE_WARNING_CONDEXPR_IS_CONSTANT
After the warning-triggering function, place this:
RESTORE_WARNINGS
Under the hood, the compiler-appropriate warning-disabling
mechanics involve the use of C99 _Pragma, which can be used
in macros.
For unrecognized or inappropriate compilers, or if
DISABLE_WARNING_PRAGMAS is defined, the macros expand
to nothing.
Reformat the list of commands, making all of them take two lines
(except for a couple that need more) even if they fit on one.
Put "#wizmgender" into alphabetical order. I don't think we need
it anymore, but if we keep the functionality then it should be
demoted from a command to a wizard mode boolean option.
Also wrap a handful of lines wider than the formatting threshold.
