Change the sortloot option to use qsort() instead of naive insertion
sort. After sorting, it reorders the linked list into the sorted
order, so might have some subtle change(s) in behavior since that
wasn't done before.
pickup.c includes some formatting cleanup.
modified:
include/extern.h, hack.h, obj.h
src/do.c, do_wear.c, end.c, invent.c, pickup.c
I upgraded from OSX 10.5.8 via 10.6.3 to 10.6.8, plus Xcode to whatever
version was on the 10.6 dvd, and ended up with a more recent version of
gcc that is configured to use 64 bit longs and 64 bit pointers (by
default; presumably that can be changed if necessary). It triggered
several warnings about converting int to pointer of different size or
vice versa even when explicit casts were in use, and a couple of other
things.
After the recent shopkeeper fix, I wanted to find out what happens if
you turn to stone in the spot inside the shop door. It didn't go too
well--a change of mine from three weeks ago caused a crash due to
passing a null pointer to strcmp(). Death from being turned to stone
or from starvation when there was no while-helpless reason (probably
not possible for starving) triggered it. This fixes that.
As far as the test goes, the shopkeeper takes your inventory and moves
it all the way into the shop, and a statue of the petrified hero is
left without contents in the spot in front of the door. That shk was
awfully quick....
Post-3.6.0 bug, so no fixes entry.
"Petrified by <foo>, while getting stoned." -- multi_reason "while
getting stoned" explains why no last-second recovery could be made,
but doesn't explain how the petrification happened, so suppress it.
"Died of starvation, while fainted from lack of food." -- nethack
does not display this; presumeably the IRC death notices for NAO are
generated from xlogfile entries. Change 'while fainted from lack of
food' to 'while fainted' at time of death if reason for death is
starvation. The longer version is accurate but sounds fairly silly.
When starvation is set in motion, set it up before checking whether
the initial faint triggers falling on a wielded cockatrice corpse, so
that fainting isn't applied after recovery in case of life-saving.
Avoid the possibility of a user-supplied name interfering with killer
reason truncation. A monster named ", while" that killed the hero
would result in "killed by <mon-type> called " being displayed on the
tombstone after stripping while-helpless reason to shorten the text.
A couple of reports asked what weird unit of measure was used for the
'realtime' value in xlogfile. It was just seconds, but was accumulating
incorrectly whenever game-state got saved for the checkpoint option.
Now it really is seconds, or rather whatever unit you get for the delta
of two time_t values; usually seconds but not guaranteed to be that.
Preformat SYSCF entry 'WIZARDS' so that it can be displayed during panic
feedback without allocating memory for the formatted list at that time.
It also gets displayed for help's "support information" ('?k').
For panic(), push "it may be possible to rebuild" to a second line since
the formatted usernames might make the line long.
Mark panic() as never returning so that code analysis might be able
to do a smarter job. It required splitting done() into two routines
since the first part really can return (but not if PANICKED was the
reason it got called). done() is now much shorter and ends with a
call to new really_done(), and panic() skips done()'s might-return
part by calling really_done() directly.
Noticed in passing: the "report error to <list of SYSCF WIZARDS>"
code calls a routine which uses alloc(), which won't work very well
if the reason for panic was because malloc() ran out of memory.
Fix some more of the complaints from clang's static analyzer. The one
in options.c (manipulating warnings symbols) appears to be an actual bug.
All the rest are either because the analysis isn't quite sophicated
enough or outright bogus.
Two of them appear to be because a static routine is attempting to guard
against callers in the same file failing to pass in required output
pointers. Stripping away the check for missing pointer should convince
the analyzer that those output parameters always receive a value. We'll
see once the analysis is eventually re-run....
exit_nhwindows() is called before terminate(), and the tty incarnation
destroys all windows--including 'pickinv_cache_win'--without setting
the various index variables used to access them to WIN_ERR, then
terminate() calls freedynamicdata() which calls free_pickinv_cache()
which tries to destroy 'pickinv_cache_win' since it isn't WIN_ERR (if
the perm_invent option has been enabled during that playing session).
Some of the other <interface>_exit_nhwindows() also tear things down
without resetting the variables used to track them, so fixing this in
exit_nhwindows() would have been pretty messy.
Call free_pickinv_cache() before exit_nhwindows() in done(). At the
moment it's only called from done(), so other exit paths won't release
the small chunk(s) of memory used for the alternate inventory window
(if it got created for perm_invent support).
This should address the issue that the problem patch to display_pickinv()
was trying to deal with: releasing the inventory window before exiting
the program so Pasi's memory checker doesn't think it's a memory leak.
Not related, but in the same file:
The older qsort comparison routines are tagged with CFDECLSPEC to deal
with some C vs C++ interaction issue. I added that to the relatively
recently added 'sortloot' qsort compare callback.
I also changed worn_wield_only(), although it isn't actually called.
(display_minventory() has provisions to call it, but both of the latter's
callers pass in MINV_ALL so allow_all() gets used instead.)
Betatesters noticed the death message could overflow the text tombstone,
because the helpless reason was also shown there.
So don't show the helpless at all in the tombstone. The helpless is
still put in the record, logfile, and as a separate while-field in xlogfile.
Make the variadic functions look more like ordinary code rather than
have the function opening brace be hidden inside the VA_DECL() macro.
That brace is still there, but VA_DECL() now needs to be followed by
a visible brace (which introduces a nested block rather than the
start of the funciton). VA_END() now provides a hidden closing brace
to end the nested block, and the existing closing brace still matches
the one in VA_DECL().
Sample usage:
void foo VA_DECL(int, arg) --macro expansion has a hidden opening brace
{ --new, explicit opening brace (actually introduces a nested block)
VA_START(bar);
...code for foo...
VA_END(); --expansion now provides a closing brace for the nested block
} --existing closing brace, still pairs with the hidden one in VA_DECL()
This should help if/when another round of reformatting ever takes place,
and also with editors or other tools that do brace/bracket/parenthesis
matching.
I had forgotten that there were variadic functions in sys/* and ended
up modifying a lot more files than intended. The majority of changes
to those just inserted a new '{' line so that revised VA_END()'s '}'
won't introduce a syntax error. A couple of them needed VA_END() moved
so that local variables wouldn't go out of scope too soon. Only the
Unix ones have been tested.
I'll push a formatting guide at some point. There may still be
outstanding changes, but please feel free to resolve those as you arrive
a them.
To the best of my knowledge, there is no changes to the actual code
content, but the formatter does have the occasional bug. If you run into
an issue, please fix it!
Replace most uses of isspace() with a simple test for ' ' after
processing the string buffer with mungspaces (which replaces tab
with space, converts instances of consecutive whitespace into a
single space, and removes leading and trailing spaces). The uses
where this wasn't done now cast their argument to (uchar) so that
platforms with signed chars will never pass negative values to it.
I didn't mess with the menu coloring code (except for casts to the
isspace() argument); it almost certainly could benefit from using
mungspaces. I did mess with the symset processing quite a bit,
and hope I haven't accidentally broken anything. Default symbols
and DECgraphics symbols still parse and display ok, so the rest of
dat/symbols should be ok too. I didn't test symbols in the user's
config file because I don't remember how that's supposed to work.
Adds the "sortloot" compound option, with possible values
of "none", "loot", or "full". It controls the sorting of
item pickup lists for inventory and looting.
Instead of just "while helpless", the death reason will tell
more explicitly why the player was helpless. For example:
"while frozen by a monster's gaze"
Add 'o' to "i a v g c" disclosure set, to display final dungeon
overview at end of game. It lists all levels visited rather than just
those that #overview considers to be interesting, but it doesn't reveal
any undiscovered aspects of those levels except for the presence of bones.
(I think revealing shops and altars and such would be worthwhile, but the
data for that isn't handy at the time.) If the game ends due to death,
the bones section of the current level will have "you, <reason you died>"
(before any real bones entries for that level). That occurs before bones
file creation so it doesn't give away whether bones are being saved.
end.c includes some unrelated lint cleanup.
Guidebook.{mn,tex} updates the section for autopickup_exceptions as
well as for disclose. It had some odd looking indentation due to various
explicit paragraph breaks. I took "experimental" out of its description
since it was moved out of the experimental section of config.h long ago.
The revised Guidebook.tex is untested.
(This covers some thing that Pat found and some things I found while working
on those.)
Unscramble duplicate use of GREPPATH and GDBPATH symbols.
Add some more info to config.h.
Make missing SYSCF_FILE a fatal error.
Make a parse error in SYSCF_FILE a fatal error.
Rename PANICTRACE_GLIBC (et al) to PANICTRACE_LIBC (et al) since FreeBSD
and Mac OS X (at least) also implement the needed API.
Allow SYSCF_FILE to be unreadable by the user (for setgid installs).
If SYSCF, do NOT fall back to the compiled in WIZARD account.
Put WIZARD into sysopt and remove special cases in authorize_wizard_mode().
Part II of the bones tracking patch. Change umpteen different outrip()
routines to handle its new time_t argument, and use formatkiller() instead
of directly accessing killer.{format,name} and killed_by_prefix[]. The
latter is now static within formatkiller().
The many sys/* and win/* changes are untested....
[See cvs log for include/rm.h or doc/window.doc for more complete description.]
Attach hero info, death reason, and date+time to a level that's being saved
as bones. Read such data back when loading a bones file, then treat it as
part of that level for the rest of the game. Dying on a loaded bones file
will chain the new hero+death+date to previous one(s) if new bones get saved.
outrip() now takes an extra argument of type time_t, and interface-specific
implementations of this routine need to be updated to handle that.
Simplify many of the intrinsics macros from
#define xxx_resistance (Hxxx || Exxx || resists_xxx(&youmonst))
down to
#define xxx_resistance (Hxxx || Exxx)
by setting or clearing an extra bit in Hxxx during polymorph so that the
resists_xxx() check becomes implicit.
Unfornately there were lots of places in the code that treat Hxxx
as a timeout number--primarily for Stunned, Confused, and Hallucination;
Stunned happens to be one of the revised macros--rather than as a bit
mask, so this patch needed a lot more changes than originally antipated.
Move the toptenwin option from flags to iflags to keep it out of
save files, thus preventing odd behavior from win32 (nethackW.exe) when
restoring and finishing games started and saved with tty (nethack.exe).
[See cvs log entry for flag.h for more complete explanation.]
The preliminary implementation of PANICTRACE on VMS had a "Fixme"
that this fixes, and a "TODO" that this makes moot, but the main reason
for this patch is that vmsmisc.c had been changed to call vms_define(),
which resides in vmsunix.c. Since vmsmisc.obj is linked into progarms
in util/ and vmsunix.obj isn't, enabling PANICTRACE caused linking
problems for those. This moves the code that wants to call vms_define()
into vmsunix.c (despite the fact that it's not even vaguely related to
Unix emulation), so that it only matters to nethack and doesn't impact
the utility programs anymore.
This uses a VMS facility called LIB$INITIALIZE to call code before
main() starts. It's rather messy--at least when written in something
other than assembler or Bliss--and shouldn't be needed for nethack,
but I couldn't figure out how to trap the condition signalled by
lib$signal(SS$_DEBUG) when the debugger isn't available to do so, so I
needed a way to make issuing that signal be conditional upon debugger
availability. One of the arguments passed to LIB$INITIALIZE-invoked
routines contains information that makes if feasible to deduce whether
the debugger is available.
Even when PANICTRACE is disabled, that's useful for handling abort
due to panic while in running in wizard mode.
A couple of extensions to the paranoid_confirmation option:
1) add paranoid_confirmation:Confirm -- setting this means that any
prompt where the other paranoid_confirm flags have been set to require
a yes response instead of y to confirm also require explicit no rather
than arbitrary non-yes to reject. It will reprompt if you don't answer
"yes" or "no" (unless you use ESC, which is treated the same as "no").
2) add paranoid_confirmation:bones -- control whether the "save bones?"
prompt in wizard mode requires yes instead of just y. The original user-
developed paranoid_confirm patch required yes unconditionally here, and
I left that out thinking it was undesireable. But after testing the
"your body rises from the dead as <undead>..." fix a couple of days ago,
where you now get an extra message and consequent --More-- prompt just
before "save bones?", I've changed my mind about its usefulness, provided
that it's settable rather than unconditional.
Handling paranoid_confirmation:bones outside of wizard mode is a
bit tricky. Right now, it can still be seen via 'O' if it has been set
in NETHACKOPTIONS, but it won't show up in the menu if you use 'O' to
interactively change the value of paranoid_confirmation. I'm not sure
whether that's the right way to go; it might be better to let non-wizard
users uselessly toggle it on and off rather than only partially hide it.
Or maybe it should be hidden from the current value even when it's set.
Or decline to set it in first place, despite external option settings.
I don't think this is useful enough to recommend ordinary users
enable it, but it's close enough to being useful that I don't want
to leave it to become subject to bit rot like umpteen other unfinished
patches. Anyone running in wizard mode who has a panic already gets
pushed into the debugger on VMS, although it doesn't work for what might
be considered the most important configuration (a secure playground, as
opposed to the wide-open one I've always been content to leave mine at).
From a bug report, receiving the
message "Your body rises from the dead as an <undead>..." gives away
the fact that bones are being created (and its absence when applicable
undead kills the hero gives away the fact that bones aren't being
created). Not very interesting for single player installations where
5-10 seconds later the player is going to check the playground for new
files, but matters on multi-user installations where players don't have
access to the directory and sometimes race each other to juicy bones,
such as nethack.alt.org.
At the end of disclosure, give the message whether bones are being
saved or not (for cases where it would have happened when bones are
created). Player won't know whether new bones are becoming available.
Also, prevent risen undead-from-hero from being given random monster
inventory, but explicitly give mummy-from-hero a mummy wrapping if the
hero isn't already carrying one. It will end up being worn; that's
the only armor mummies are allowed to put on.
[Short writeup; see 'cvs log' of flag.h or options.c for the long one.]
This is a reworking of user contributed patch known as Paranoid_Quit.
Add a new compound option, paranoid_confirmation, accepting a space
separated list of values "quit die attack pray Remove"; default is "pray".
paranoid:quit - yes vs y for "really quit?" and "enter explore mode?"
paranoid:die - yes vs y for "die?" in explore mode or wizard mode
paranoid:attack - yes vs y for "really attack <peacful monster>?"
paranoid:pray - y to pray; supersedes prayconfirm boolean; on by default
paranoid:Remove - always issue an inventory prompt for 'R' an 'T', even
when only one applicable item is currently worn.
Rename ``kickobj'' to ``kickedobj'' so that the tense matches that
of ``thrownobj''. Also, move their declarations to decl.h and their
definitions to decl.c since usage has spread from dokick.c/dothrow.c to
various files and is about to expand to another one.
This is all tiny stuff - allow overriding WIDENED_PROTOTYPES from the hints
file, missing NO_SIGNAL conditionals, remove a GCC-ism, conditional indentation,
void return in a non-void function.
Add missing check for NO_SIGNAL in PANICTRACE versionof NH_abort().
Fix indentation of pre-processing directives.
Add #endif comments to make things clearer.
On crash signal or panic(), use a configurable method to get a stacktrace
the user can easily report to us. Currently only for Unix/Linux and only
ifdef BETA. Hopefully ports can add additional methods.
Bits:
- linux hints file had PREFIX definition in the wrong place
- sample sysconf file used wrong delimiter for WIZARDS
- fix grammar error in support message when using sysconf.wizards
- options.c comment typo
- capitalize "Crash test" output from #panic command
