| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| |
|
|
|
|
|
| |
.cc, moving its contents into the new stepdown.cc and strings.cc.
(The latter also got many donations from libutil.h.)
Down with stuff! Up the new flesh!
|
| | |
|
| |
|
|
|
| |
Remove traditional crosstraining.
Replace it with a 20% skillpoint enhancer bonus.
|
| |
|
|
|
|
|
|
| |
We made the decision a while ago to remove anti-training.
This had come up in discussions about changing
crosstraining and it seemed that no one actually likes
anti-training. Rather than blocking on cross-training,
let's just put this in.
|
| |
|
|
|
| |
Some of them were missing the /** (or /*!) that makes doxygen consider
the comment in the first place.
|
| |
|
|
| |
Delete train_skill, an unused function.
|
| |
|
|
| |
Since the card was removed and it only remained as a Zotdef effect.
|
| |
|
|
|
|
| |
For Shields in particular, carrying a shield is sufficient for training
purposes. Remove a number of now-unnecessary checks, including some checks
that were already obsolete for curses on wielded weapons.
|
| | |
|
| |
|
|
| |
The old name was quite puzzling...
|
| |
|
|
|
| |
Sometimes, they're there to emphasize a break between two sections of code,
which is good. In a majority of cases, though, they're just inconsistent.
|
| |
|
|
|
| |
The translation project is stalled, but this makes all of such static strings
trivially gatherable without any extra work.
|
| | |
|
| |
|
|
|
| |
map or set.count() can test the presence of a given key and return 0 or 1
outright.
|
| |
|
|
| |
Also, simplify an expression such removal of parentheses revealed.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This should hopefully make games from the same seed proceed the same on all
architectures[1], as I can't think of anything else that behaves differently
based on pointer size, endianness or exact pointer values.
I used a NIH implementation instead of passing a third argument to
random_shuffle, as the interface is so much nicer.
[1]. This will be affected by terminal size (elemental colours are not
resolved outside the screen), tile windows size (random animations),
tiles/non-tiles and possibly others.
Also, I don't think we use STL hashes anywhere, but if we'd do, the STL
implementation will matter. Please don't make this stop you: this commit
is only to make test cases from my stress tests portable which is a small
benefit, perhaps even smaller than the nicer call interface.
|
| | |
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
As a second facet of the attempt to reduce player power growth into
mid and lategame, the xp requirement of experience levels above 15
is steadily increased, with xl 16 requiring 3% more xp than previously
while xl 27 requires 25% (xl 13-15 cost slightly less than before).
Skill point costs are also increased in tandem, though the reduction
in skill growth should be somewhat less than the one in xl.
This also smooths out skill gain in the mid-teens, where the old values
led to a sudden spike of growth which then sharply trailed off shortly
after Lair (or thereabouts). The rate of change in growth should hopefully
feel more continuous, instead of a noticable stall as one proceeds into
midgame.
|
| |
|
|
| |
It's put back into an int variable right after, so range is not an issue.
|
| |
|
|
|
|
|
|
|
|
|
| |
We use GCC version <= 4.1 as the criterion for the fix (which is the
same as the similar Solaris fix). Probably we could use __GLIBCXX__
and/or __GLIBCPP__ to detect this better, but I haven't been able
to test that (in particular, old g++ with a different libstdc++).
This reverts commit 0e1e71dfe8e254a2d925fcb0a7ee6905c4175dd1. We still
have to do some kind of casting for that particular case, though,
because the first argument is unsigned.
|
| |
|
|
|
|
| |
Committing separately as I'm not sure whether checking, for example,
ASSERT_RANGE(level, 1, 28) is that nice. Perhaps 27 + 1 could be better?
Perhaps some other syntax?
|
| |
|
|
|
|
| |
Multiple manuals for different skills can be used at once. If there are
multiple manuals for the same skill, the one with least charges is used
up first.
|
| |
|
|
| |
These are in the standard but break on MSVC.
|
| | |
|
| |
|
|
|
| |
The only cost is having assignments be a function, as you can't overload
operator[] to have separate bits be lvalues.
|
| |
|
|
|
|
|
|
|
|
|
|
| |
Per one of proposals, its effect is now an average of the weapon skill
and Stealth. Stabbing tiers remain as before:
* daggers
* other short blades
* long blades, spears, small club
* other stuff
Let's see how this works. We can always change the formulas, and at
least the skill is gone.
|
| |
|
|
|
|
|
|
|
|
|
| |
While this message cannot be totally ignored, as it warns about wasting xp,
usually it means you could potentially learn a spell you don't want just to
place xp in its skill. That firestorming conjurer may have better things
to do than lugging a weapon of yet another type, or learning a cold spell.
I made this message repeat itself:
* when you reload the game
* after exiting the 'm' screen
|
| |
|
|
|
|
| |
Other mode being either auto or manual depending on which one is active.
Probably the cause of #6507. It's only a display bug by the way, level 27
skills are never trained.
|
| |
|
|
|
|
|
|
| |
Reading a (non-useless) manual enabled training for its skill, even if
training would otherwise be blocked (e.g. a manual of Shields without a
shield, or a manual of Ice Magic with no relevant spells). However, on
save and reload the skill was marked as untrainable. Fix this by always
removing the player's manual_skill from stop_train.
|
| |
|
|
|
| |
It was a remnant of the old system in which raising skills to 1 meant
something special, and it was insignificant anyway.
|
| |\
| |
| |
| | |
Yay for conflicts galore (wax removal, std:: purge, tileidx changes).
|
| | |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
It is now consistent with the skill apts. Races with a -1 apt require 18% more
experience to reach a given XL than races with a +0 apt do.
The old apts were converted to the new scale as follows:
100 -> +1
110,120 -> 0
130,140,150 -> -1
160 -> -2
|
| | |
| |
| |
| |
| |
| |
| |
| |
| | |
We were getting an ambiguous overload with between pow(double, double)
and pow(double, int) when calling pow() with two ints. It didn't show
up before 0.12-a0-208-g770bcbd because the version taking an int was
hidden in the std:: namespace.
This doesn't seem to be a problem with GCC 4.6.
|
| | |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
I had to rename distance() (in coord.h) to distance2() because it conflicts
with the STL function to compare 2 iterators. Not a bad change given how it
returns the square of the distance anyway.
I also had to rename the message global variable (in message.cc) to buffer.
I tried to fix and improve the coding style has much as I could, but I
probably missed a few given how huge and tedious it is.
I also didn't touch crawl-gdb.py, and the stuff in prebuilt, rltiles/tool
and util/levcomp.*, because I have no clue about those.
|
| | |
| |
| |
| |
| | |
On top of getting negative rot, you could also scum for free maxMP by
toggling spellcasting on/off.
|
| | |
| |
| |
| |
| |
| |
| |
| | |
Also, remove a few special cases that considered Spellcasting to be
non-magic below level one for Troglodytes; this hasn't been necessary
since scrolls stopped training Spellcasting in ee84c39.
This reverts commit 1d240c20e937d950ff3de459d85665fcc06969ff.
|
| | | |
|
| | |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
The only popular one where this works is amd64, and even that reached 50%
use in Linux on desktops/laptops only recently. Linux i386, Windows, arm*,
etc, are still 32 bit, so are many less popular platforms.
Nearly any use of "long" in portable code is a bug, except for some things
where using a full machine word improves efficiency but you can still handle
shorter words properly. Elsewhere else, either an int is enough (and you'd
waste memory on 64 bit), or int is not enough and it will overflow on 32 bit.
|
| | |
| |
| |
| |
| |
| | |
The skill exercise arrays would suffer integer overflow during the
re-initialisation routines; this adjusts the relevant code to use longs
(shudder) to avoid that issue.
|
| | |
| |
| |
| |
| | |
The previous opposite order was confusing the skills screen on a
cursed or distortion item wield (0005920).
|
| |/
|
|
|
| |
The only relevant use left, silently opening doors, is now 100% Stealth
(and DEX), rather than 50% Stealth 50% T&D.
|
| | |
|
| | |
|
| |
|
|
|
| |
When all your skills are either maxxed or untrainable. Just print the message
once when the last trainable skill gets disabled.
|
| |
|
|
|
|
|
|
|
| |
These accumulate but never get removed; no wonder compilation times keep
rising.
The includes.sh script has lots of false negatives (and positives...), and
can't check .h files which cause the biggest slowdown, it'd be nice to run
multidelta on those somehow.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
You could tediously id anything by bashing a plant or throwing it at a wall,
with as little as 0.1 skill. This is no more: every weapon has a secret
skill level needed to auto-id it. You must know enough about the weapon to
possibly id it under the old system -- having wielded it once is enough, so
are certain other ways. I intentionally didn't go into the complexity of
checking current curse status or blocking distortion/vamp.
Save compat: on training a weapon skill and certain other events for the
first time, you'll get unsightly ids of a bunch of items. Probably not
worth any effort, you are very unlikely to get multiple ids at once during
normal play.
This commit does not handle missiles: they are mergeable and thus don't have
an unique random number, it would be enough to id one with the smallest
number and then merge stacks, you are currently encouraged to keep one of
every missile with different enchantments, etc.
|
| | |
|
| | |
|
| | |
|
