If a punished player picks up the iron ball, gets engulfed and
saves, then the saved game will have missed saving the dangling
chain since it was not on the floor or in the inventory. Upon
restoring the saved game, the game will be in a bad state since
the ball will be worn but the chain will be missing.
Make some progress on a couple of next minor release checklist
items, hopefully without introducing too many new bugs. This
is just the initial commit, and work continues.
Checklist items:
Savefiles compatible between Windows versions, whether 64-bit
or 32-bit in little-endian field format.
Selection of file formats:
historical (structlevel saves),
lendian (little-endian, fieldlevel saves),
and just for proof-of-concept, ascii fieldlevel saves
(the ascii is huge! 10x bigger than little-endian).
For the fieldlevel save, all complex data structures recursively
get broken down until until it is one of the simple types that
can't be broken down any further, and that gets when it gets
written to the output file.
New files needed for this build:
hand-coded:
include/sfprocs.h
src/sfbase.c - really a dispatcher to one of the
output/input format routines.
src/sflendian.c - little-endian output writer/reader.
src/sfascii.c - ascii text output writer/reader.
auto-coded (generated):
include/sfproto.h
src/sfdata.c
This is just one approach. I'm sure there are countless others
and they have different pros and cons.
For producing the auto-coded files a utility called
universal-ctags, that is actively maintained and evolving,
was used to do all the heavy-lifting of parsing the
NetHack C sources to tabulate the data fields, and store
them in an intermediate file called util/nethack.tags
(not required for building NetHack if you already have a
generated include/sfproto.h and src/sfdata.c)
util/readtags (also not required for building NetHack
itself) will decipher the nethack.tags file and produce
the functions that can deal with the NetHack struct data
fields.
You can obtain the source for universal-ctags by cloning it
from here:
https://github.com/universal-ctags/ctags.git
The combination universal-ctags + util/readtags has been
tried and tested under both Windows and Linux, so it is
not tied to a particular platform.
Note: util/readtags will work only with universal-ctags
output, so other ctags are unlikely to work as-is.
Universal-ctags can be build from source very easily
under Linux, or under Windows using visual studio.
This doesn't solve the <0,0> problem but it does prevent mnexto()
from using uninitialized coordinates if enexto() fails. It also adds
several debugging messages.
enexto() was ignoring map row #0 (unlike column #0, row #0 contains
valid map locations). Fixing that doesn't matter for Plane of Water
though since that row is stone there--that's probably a bug. It was
also repeatedly re-testing the top+1 and bottom rows and left and
right columns after they had already failed to be acceptable. It
still does some of that, but less.
Extend 'putstr(WIN_MESSAGE, attribute, string)'s attribute so that
'custompline(SUPPRESS_HISTORY, ...)' can work with ^P's message
history like DUMPLOG history, in order to keep autodescribe feedback
and intermediate prompts for multi-digit count ('Count: 12', 'Count:
123') prompts out of recall history. The old autodescribe behavior
could easily push all real messages out of the recall buffer when
moving the cursor around for getpos, and the count behavior looked
silly for a four or five digit gold count if you set the msg_window
option to 'full' or 'combination' and viewed them all at once.
Other interfaces may want to follow suit, but this doesn't force them
to make any changes. I added a hook for "urgent messages" that might
be rendered in bold or red or some such and/or override the use of
ESC at --More-- from suppressing further messages, but there aren't
any custompline(URGENT_MESSAGE, ...) calls (potentially "You die...",
for instance) to exercise it. Other people have implemented similar
feature it different ways and I'm not sure whether this one is really
the way to go since the core needs to categorize each message that it
deems to be urgent. MSG_TYPE:stop may be sufficent, although MSG_TYPE
matching can entail a lot of regexp execution overhead at run-time.
add MM_NOGRP makemon() flag as a means of suppressing groups of monsters in
a couple places that warrant it when a specific monster type isn't
specified on the call to makemon()
Force trap to activate during failed untrap attempt if done while
already at the trap's location, to match the recent change in
behavior when failed attempt occurs while adjacent to the trap.
Most noticeable while flying over bear traps, but affects all
failed untrap attempts.
Changes to be committed:
modified: include/decl.h
modified: include/dungeon.h
modified: include/extern.h
modified: include/hack.h
modified: src/decl.c
modified: src/do_name.c
modified: src/dog.c
modified: src/dokick.c
modified: src/makemon.c
modified: src/mkmaze.c
modified: src/mkobj.c
modified: src/pager.c
This commit is an attempt to address the complaints about
the orc town variation taking away lots of stuff that is
normally available in mine town. The statement in the level
description says "A tragic accident has occurred in Frontier
Town...It has been overrun by orcs."
The changes in this commit attempt to uphold that premise,
while making things a bit more interesting and perhaps
more palatable for the player.
This update does the following in keeping with the mythos:
- While many of the orcs still remain to wander about the
level, many of the orcs took off deeper into the mines with
some of the stuff that they plundered. You may now be
able to hunt some of it down.
- Adds some appearance of this particular horde of marauding
orcs working as part of a larger collective.
- This evolves the Orc Town mine town variation into a
a feature over multiple levels of The Gnomish Mines,
rather than just the single-level "feature" that it was
previously.
- You may have to work longer and a bit harder for some
things than other mine town variations, but at least with
these changes, there is hope that some of it may be found
elsewhere.
Game mechanics notes (maybe spoily?)
- Add mechanism to place objects into limbo (okay, really
place them onto the migrating_objs list for transferring
between levels etc.) and destine them
to become part of the monster inventory of a particular
species. In this particular usage case, it's using the
M2_ORC flag setting to identify the recipients.
- At present, there is no mechanism in the level compiler
for placing objects onto the migrating objects, nor
with more sophisticated landing logic, so a somewhat
kludgy hard-coded fixup and supporting routines were used.
Some day the need for that might change if additional
capabilities move to the level compiler.
This is a NetHack-3.6.2-beta01 update. Please give it a workout.
Fixes#127
Build fix, avoid use of 'class'
include\hack.h(199): error C2236: unexpected token 'class'. Did you forget a ';'?
include\hack.h(199): error C2332: 'class': missing tag name
include\hack.h(199): error C2027: use of undefined type 'sortloot_item::<unnamed-tag>'
When objects are in the same class, sortloot orders them by their
formatted name. It was reformatting each object every time it got
compared to another object. Change that to remember the formatted
name so that any given object is formatted at most once (during the
current sort; future sorts will need to format it again).
Armor and weapon classes are subdivided into smaller subclasses
and the formatting plus alpha compare is only done for items in
the same subclass, so helms come out before cloaks and don't get
their names compared, for instance. [That was from my 'revamp'
rather than the original implementation.] This adds a couple more
subclass sets: food (named fruit, 'other' food, tins, eggs, corpses,
globs) and tools (containers, pseudo-containers [bag of tricks and
horn of plenty once those have become discovered; prior to discovery,
bag of tricks is classified as a container and horn of plenty as an
instrument], instruments, 'other' tools).
The main difference, aside from the formatting efficiency improvement,
is to change the previous sort order
| pink potion
| potion of enlightenment
| purple-red potion
to be
| pink potion
| purple-red potion
| potion of enlightenment
by grouping undiscovered items before discovered items when class and
subclass match. So discovery state is essentially a sub-subclass and
formatting plus string comparison is only done for members of the
same sub-subclass. There are actually four state values: unseen
(which applies to particular objects rather than to their type),
unknown (not discovered and not named), named (not discovered but has
player-assigned type name), and discovered (either fully discovered
or considered not interesting to discover [no alternate description,
not nameable]).
My testing was primarily done with pickup ('m,' with menustyle:T)
and sortloot:Loot (the default) plus !sortpack (not the default and
not a setting I ordinarily use, but less verbose without the class
separators). It won't astonish me if oddities crop up with other
usage combinations.
Yesterday's sortloot() overhaul didn't include some cockatrice corpse
handling for pickup. If there's an object class filter in place and
pickup has been told to care about cockatrice corpses, have sortloot()
include them in the loot array even if food class isn't accepted by
the filter. In the pre-sortloot days, and in 3.6.[01] which didn't
attempt to deliver a filtered subset of loot, the check for such
corpses was done before pickup checks the filter. They need to be in
the loot array to retain the same behavior.
H7205 - full-pack identify might skip items if perm_invent is on
because updating the inventory window might reorder 'invent'
while the identify code is in the midst of traversing it;
H7120 - pickup that doesn't pick anything up can change the glyph
shown on the map because the pile might be reordered such
that a different item is on top;
H5216 - performing a sortloot operation on a pile and then switching
back to sortloot:none doesn't restore pile's original order.
The 'revamp' that changed the contributed sortloot feature to switch
to simpler usage (object list itself was sorted rather than having a
parallel array that needed to be constructed, sorted, traversed, and
discarded) turns out to have too many problems. This reverts to a
hybrid solution that constructs an array for traversal, leaving the
linked list in its original order, but hides most of the details of
that from sortloot() callers. The 'revamp' benefit of being able to
use normal list traversal is lost, as is the potential to skip
sorting when the list turns out to already be in the desired order.
This could stand to have a lot more testing than it's had so far.
Defined strbuf_t and related routines to support dynamically sized
strings. Modified strip_newline() to strip the last newline in a string
instead of the first.
Simplified splash window code using new strbuf_t.
Prior to exiting game, re-enable getreturn and call wait_synch() in
case there is buffered raw prints that must be displayed to user.
The different menustyle settings have been offering different degrees
of support for BUCX filtering:
Full : multi-drop, container-in, container-out
Trad, Combo: multi-drop
Partial : none (to be expected; it explicitly jumps past class
filtering, where BUCX gets handled, to a menu of all objects)
This adds pickup, container-in, container-out, multi-unwear/unwield,
and object-ID for Trad and Combo, and multi-unwear/unwield for Full.
(Full behaves like Partial for pickup--not sure why--and for object-ID,
bypassing filters to go straight to a menu of all applicable items.)
There are probably several new bugs--this stuff is very convoluted....
I couldn't reproduce the reported problem of the "In what direction?"
being issued after the screen was cleared, but bypassing pline() in
favor of putstr(WIN_MESSAGE) for tty prompts did also bypass
if (vision_full_recalc) vision_recalc(0);
if (u.ux) flush_screen(1);
done in pline(). Inadvertent loss of the latter could conceivably be
responsible for the problem. If so, the escape code used by cl_end()
may be broken for somebody's termcap or terminfo setup since clearing
to the end of the line in the message window shouldn't erase the rest
of the screen.
Regardless, the prompting change also bypassed the ability to show
the prompt with raw_printf() if the display wasn't fully intialized
yet, so some change to the revised prompting was necessary anyway.
Switching back from putstr(WIN_MESSAGE) to pline() resulted in
duplicated entries in DUMPLOG message history, one with bare prompt
followed by another with response appended, so more tweaking was
needed. The result is use of new custompline() instead of normal
pline(). custompline() accepts some message handling flags to give
more control over pline()'s behavior. It's a more general variation
of Norep() but its caller needs to specify an extra argument.
Add MG_BW_LAVA to mapglyph() instead of hijacking MG_DETECT. Used
to display lava in inverse video if color is disabled and lava is
using the same display character as water (which is the default).
(The use_inverse option must be enabled for tty to honor it. X11's
text mode doesn't care. Win32 does care but probably shouldn't--it's
not a case like tty where the hardware might not support it.)
This implements both MG_DETECT and MG_BW_LAVA for X11, but only if
the program is built with TEXTCOLOR enabled. Those should work even
when color is not supported, but I suspect that configuration is
unlikely to ever be used so didn't want to spend the time to figure
out how to do it. (The relevant data is overloaded on the color
data, so not available when TEXTCOLOR is disabled.)
The win32 revision is untested.
Do it properly, using the arguments to xkilled() instead of reversing
the conduct counter after the fact.
The xkilled() flag value of '1' has been reversed. It used to mean
'display message' but now means 'suppress message' since both of the
other flag bits are for suppression. All callers have been updated
to specify either XKILL_GIVEMSG or XKILL_NOMSG so the underlying
number remains transparent.
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
Test case: U-shaped corridor, with a known trap in it.
Before this change, travel would try to move straight at
the target, bumping the wall or walking into a dead-end.
After this, travel will go along the corridor and then stop
right before the trap.
Original patch via AceHack by Alex Smith.
Add MG_OBJPILE flag, which windowports can use to check if a location
has more than one object stack. If use_inverse is on, TTY will use
inverse to show such piles. If a boulder is the topmost item on a pile,
then the object pile flag is not used; mainly because boulders are "solid",
boulders dropped by monsters are nearly always over other objects, and so
that special levels such a Sokoban can "hide" items under the boulders.
TODO: a "pilemark", analogous to "petmark", perhaps a green plus sign,
which can be used by windowports with tiles.
