<Someone> noticed that if one builds something in util and the required .o
files aren't already built in src, the .o gets placed in the wrong place.
Added the missing '-o $@' to the compilation command.
Reported on RH 7.2 and 8.0. Compilation failed because system headers that
needed _GNU_SOURCE on these Redhat versions got included before it was
defined. To ensure _GNU_SOURCE is defined, added an autodetect for it to
config1.h and removed the need to set it in unixres.c. __linux__ is also
checked elsewhere.
As reported, if you're twoweaponing and die, your secondary weapon may
become cursed and drop. But, the bones file code is dropping everything
and tries to drop it again, causing a panic. drop_upon_death just clears
things out, so follow suit for uswapwep.
I'm updating the official binary to be built on a Redhat 8.0 system. This
should give the binary some legs if we take a while to get the next release out.
> There is a slight problem with "--more--" prompt being handled in
> WM_PAINT. WM_PAINT can be sent while you're inside
> mswin_nhgetch(). This causes some weirdness if window is resized
> while waiting for --more-- prompt.
> Why does warning work normally while hallucinating? I'd like to see
> random numbers.
It's just a one-liner, although it has pointed out problems in the
symbol lookup code. "0" won't work at present, so this doesn't use it,
but warning symbols "1" through "5" also have trouble--regardless of
hallucination--if someone overloads them to represent any feature aside
from warning.
1) Update documentation for using bison/flex
2) Fix use of $(CC). In the makefile, it's supposed to be $(cc).
Otherwise it won't compile. [caused during recent dependency
updates]
'+' is not a valid argument to resist(), although it still works
since the default case catches it. Make the spell class explicit,
although I don't think having the effectiveness of scrolls of confuse
monster depend on the character's level makes much sense.
The CE ports use makedefs hosted on another platform,
so the version string generated at build time isn't really
appropriate.
Add a way to add information to the version string
at runtime for such ports.
Reported last spring and again last week: monsters who teleport to
your location (named demons, master liches, and so forth) continue to do
that even if they've just used a scroll or wand of teleportation to get
away from you. This doesn't prevent that situation but does make it be
much less likely to occur.
The previous changed ended up discarding the begining portion of
excessively long lines and keeping the end. It's unlikely that either
part is going to be valid, but reporting the ending portion as a failed
wish would make tracking down and fixing the situation trickier.
<email deleted>
Newsgroups: rec.games.roguelike.nethack
Subject: Re: Other ways to destroy the Amulet
<Someone> wrote:
>>> Zapped undead turning at his _corpse_, which brought the corpse
>>> back to life and also destroyed the ghost. The revived
>>> character was not carrying anything.
>>
>> Hmm, didn't the Amulet fall to the floor like when you genocide
>> something that's carrying something?
>
> Nope.
>
> That's because undead turning does mongone(ghost). The mongone
> function destroys the inventory of the monster:
> discard_minvent(mdef); /* release monster's inventory */
>
> Whereas genocided monsters are removed with mondead, which calls
> m_detach, which calls relobj, which drops the monster's inventory
> on the floor.
The "following" flag could get set in several places where it was not
obvious that the customer name would be remembered. Since the shopkeeper
should always get angry at the current player, set the name at the same
place that the flag is set.
