Rewrite the broken code to move cargo with its carrier
The old code did not move a carrier's cargo (planes, land units,
nukes) when the carrier moved. Instead, it fixed up the location in
the postread callback. Anything not going through ef_read(), in
particular the update, saw it in its old, incorrect location, until a
fixed up copy got written back.
Moreover, the timestamp did not change when cargo moved, so
incremental dumps did not pick up the movement.
The new code moves the cargo along with the carrier.
New unit_update_cargo() moves or destroys a carrier's cargo (planes,
land units, nukes) along with the carrier. Call it from
shp_prewrite(), pln_prewrite() and lnd_prewrite() when the carrier
moves or gets destroyed.
Remove the code to destroy cargo from shp_prewrite(), pln_prewrite(),
lnd_prewrite().
Remove the code to fix up cargo location from pln_postread(),
lnd_postread(), nuk_postread().
This changes the message for ship and land unit cargo getting
destroyed from "sunk" and "MIA" to "lost".
Make the item iterator capable of iterating over a cargo list
New snxtitem_cargo() initializes an iterator for a cargo list, with
new enum ns_seltype member NS_GROUP and new struct nstr_item member
next. Extend nxtitem() and nxtitemp() to step through the list.
Zap next uid in clink_rem(), check it in clink_add()
The former ensures that next links are valid even for uids not on any
list. The latter oopses on adding an uid to a list when it is already
on a list, unless it is at the tail.
Ron Koenderink [Fri, 12 Sep 2008 01:06:10 +0000 (19:06 -0600)]
Ensure the empth_sleep() always yields for WIN32
The WIN32 version did not block when the sleep was
already reached by the time empth_sleep() did the
time remaining calculation. The other versions
of empth_sleep() do always yield.
Ron Koenderink [Thu, 11 Sep 2008 20:58:32 +0000 (14:58 -0600)]
Use empty schedule file for the nightly build
This ensures that no unplanned updates occur
during the nightly build sequence. Remove
unnecesary enables and disables from the script.
Remove unnecessary argument for force command.
Instead of counting the load with lnd_nland() / lnd_nxlight(), check
whether there's at least one loaded with lnd_first_on_land() /
pln_first_on_land().
Load counters are redundant; they can be computed from the carrier
uids. Keeping them up-to-date as the carriers change is a pain, and
we never got that quite complete.
Computing load counters straight from the carrier uids every time we
need them would be rather inefficient, but computing them from cargo
lists is not. So do that.
Remove the load counters: struct shpstr members shp_nplane,
shp_nchoppers, shp_nxlight, shp_nland, and struct lndstr members
lnd_nxlight and lnd_nland.
Nothing left in fit_plane_off_ship(), fit_plane_off_land(), so remove
them.
load_land_ship(), load_land_land(), check_trade(), pln_newlanding(),
put_plane_on_ship(), take_plane_off_ship(), put_plane_on_land(),
take_plane_off_land() no longer change the carrier, so don't put it.
Remove functions to recompute the load counters from carrier uids:
count_units(), lnd_count_units(), count_planes(), count_land_planes(),
pln_fixup() and lnd_fixup(), along with the latter two's private
copies of fit_plane_on_ship() and fit_plane_on_land().
New cargo list functions to compute load counts: unit_cargo_count()
and unit_nplane(), with convenience wrappers shp_nplane(),
shp_nland(), lnd_nxlight(), lnd_nland().
Use them to make ship selectors nplane, nchoppers, nxlight, nland
virtual. They now reflect what is loaded, not how the load uses the
available slots. This makes a difference when x-light planes or
choppers use plane slots.
Use them to make land unit selectors nxlight and nland virtual.
Use them to get load counts in land(), ldump(), load_plane_ship(),
load_land_ship(), load_plane_land(), load_land_land(), sdump(), shi(),
tend_land(), fit_plane_on_land(), trade_desc(), unit_list().
Rewrite fit_plane_on_ship() and could_be_on_ship() to use
shp_nplane(). could_be_on_ship() now takes load count arguments, as
computed by shp_nplane(), so it can be used for checking against an
existing load as well.
Cargo lists storing lists of cargo for each carrier
Persistent game state encodes "who carries what" by storing the
carrier uid in the cargo. Cargo lists augment that: they store lists
of cargo for each carrier. They are not persistent.
New unit_cargo_init() to compute the cargo lists from game state.
Call it in ef_init_srv() and at the end of update_main().
New unit_onresize() to resize the cargo list data structure.
Installed as units' struct empfile callback onresize to make them
resize automatically with the unit files.
New unit_carrier_change() to update cargo lists when carriers change
in game state. Convenience wrappers pln_carrier_change(),
lnd_carrier_change() and nuk_carrier_change(). Call them from
prewrite callbacks to keep cargo lists in sync with game state.
To make that work, unused units must not point to a carrier. Add new
pln_oninit(), lnd_oninit() and nuk_oninit() take care of newly created
units. Change lnd_prewrite() and nuk_prewrite() to take dead land
units and nukes off their carrier. pln_prewrite() did that already.
New unit_cargo_first(), unit_cargo_next() to traverse cargo lists.
Convenience wrappers lnd_first_on_ship(), lnd_first_on_land(),
lnd_next_on_unit(), pln_first_on_ship(), pln_first_on_land(),
pln_next_on_unit() and nuk_on_plane(). The latter is disabled for now
because it clashes with an existing function.
Move selector code from src/lib/global to src/lib/common
Future virtual selectors will need to access game state. This depends
on common/file.c, which can't be used from global without creating a
cyclic dependency between libglobal.a and libcommon.a.
Move nsc.c to src/lib/common. file.c depends on it, so move it as
well, renamed to filetable.c so it doesn't clash with the existing
file.c.
Losses of sectors, ships, planes, land units and nukes are tracked in
the lost file. To keep it current, makelost() and makenotlost() were
called whenever one of these changed owners. Cumbersome and
error-prone. In fact, the lost file was never perfectly accurate.
Detect the ownership change in the prewrite callback and call
makelost() / makenotlost() from there. Remove lost file updates from
where they're no longer needed: right before a put. takeover() is a
bit more involved: it doesn't put the sectors, but all callers do,
except for guerrilla(). So remove the lost file update from
takeover(), but add it to guerrilla().
This takes care of lost file update for all ownership changes that go
through ef_write(). It can't take care of any missing updates for
changes that don't go through it.
Don't execute pre-write checks from sct_postread()
Chainsaw 3 factored out most checks from sct_prewrite() into
checksect(), and called that from sct_postread() as well. Revert
this, because it was a bad idea: whenever checksect() called from
ef_read() found something to change, it was actually a bug, and
changing it just hid the bug from whatever called ef_read(). But it
couldn't hide it from code going through ef_ptr(). So, instead of
having bugs visible everywhere, including census and such, they were
hidden in hard to observe places. For instance, the previous commit
fixed one that was visible to the path finder, but not the actual path
user, which led to the path user choking on an incorrect path in a
rather obscure manner.
Things no longer "corrected" on read: excessive mobility, low work
despite no civilians present, sector owned despite neither civilians,
military nor land units. The latter had a scary-looking caploss(),
but NF_SACKED should have rendered that harmless.
Don't revert sectors without military to old owner
Change checksect() not to abandon occupied sectors to the old owner
when there is no military and no land units. This effectively
restores pre-Chainsaw 3 behavior. Matching change to would_abandon().
Rationale. Traditional ways to change sector owner:
(1) Attack, assault, paradrop can transfer a sector to the attacker,
in take_def().
(2) Guerrilla warfare at the update can transfer a sector to the old
owner, in guerilla(). This happens when che kill all military and
the sector is sufficiently disloyal to the owner.
(3) Whenever all civilians, military and land units are removed from a
sector, no matter how, it silently reverts to the deity, in
checksect().
Chainsaw 3 added:
(4) Whenever all military and land units are removed from an occupied
sector, no matter how, it silently reverts to the old owner, in
checksect().
This addition isn't seamless. Funnies include:
* When che kill all military and land units, but the sector is loyal,
(3) doesn't transfer to the old owner. But since there's no
military and land units left, (4) transfers it anyway, only without
telling the lucky old owner. The latter transfer is buggy:
checksect() runs only on ef_read() and ef_write(), not the update
(bug#1010856), so the silent transfer is delayed until the next
ef_write(). But code using ef_read() sees it right away. For
instance, the path finder, which doesn't use ef_read(), can route a
path through a sector lost that way. The actual move, which does
use ef_read(), then chokes on that path.
* When you attack a sector, and get defeated with the help of reacting
land units, but succeed in killing the *local* defenders, (4) makes
the sector silently revert to the old owner. Which might be
somebody who wasn't involved in the fight, and gets no notification
whatsoever of his windfall.
* You can abandon a sector to the old-owner by removing all military
and land units from it, in a myriad of ways. Some ways ask you for
confirmation (move, march, load), many don't (navigate, plane
construction, delivery; arguably bugs), and others simply don't let
you (paradrop, fly, distribution).
This problem also exists for abandoning to deity, i.e. through (3)
instead of (4). Some ways to move out civilians don't let you do
that (distribute), but most do. However, accidentally abandoning an
empty sector to deity is less serious than a populated one to
another player.
In my opinion, (4) doesn't add much to the game, and fixing the
funnies isn't worth the effort.
Empire3's C_SYNC code added these to sct_prewrite(), shp_prewrite(),
pln_prewrite(), lnd_prewrite() and nuk_prewrite(). They weren't
removed when C_SYNC was ripped out in 4.0.0.
Really belongs there, because it manipulates empfile[].
New ef_open_view() to replace ef_init_view(). Make ef_close() cope
with views, and remove ef_fina_view(). Make ef_extend() and
ef_truncate() oops on views.
ef_elt_by_name(), xdprval_sym() and symval() checked whether a file
type T is a symbol table by comparing ef_cadef(T) to symbol_ca, even
though T may be EF_BAD. Before commit 50cfdcb5, ef_cadef(EF_BAD)
accessed empfile[] out of bounds, which could conceivably crash or
somehow happen to yield symbol_ca. Since then, it oopses and returns
null.
Fix by testing the file type before calling ef_cadef().
Xundump had special hackery to maintain configuration tables'
sentinels: xubody() and getobj() added a sentinel element when
initializing or growing a table, which xubody() stripped off again
before returning. The latter was an unclean hack.
Replace this by building knowledge of sentinels into struct empfile:
new flag EFF_SENTINEL, set for the appropriate members of empfile[],
obeyed by ef_extend() and ef_truncate().
Fix/improve logging in ef_close(), ef_extend(), ef_truncate()
Change ef_close() to log ep->file instead of ep->name, to match
ef_open().
Fix ef_extend() to log ep->name instead of ep->file, which could be
null. Also fix ef_ensure_space()'s function comment. Both broken in
commit 2eb8672b.
ef_truncate()'s error logging lacked detail when ef_realloc_cache()
failed, fix.
It became needlessly complicated in 4.0.1 to fix a "bug in mapdist not
taking world edges into account nicely enough." That "fix" had no
effect, which was good, because it wasn't broken.
Include destination in interception and plane mission messages
Interception and missions launch planes automatically. The plane
owner (and for missions, the base sector owner) gets a message.
Destination coordinates are often obvious from the context, but not
always, e.g. when flying in support of allies, or when the mission
gets interepted and aborts. Include the destination coordinates in
the messages. Reported by Ulrich Hannemann.
Fix client code for catching rogue redirections and executes
The client permits redirection and execute only if it recently saw
them in player input. The caching of recent player input for this
purpose was broken.
When save_input() evicted a line, it kept its newline. This could
make the recent_input ring buffer fill up, which then got redirections
and execute misdiagnosed as rogue.
save_input() also screwed up when evicting one line didn't make
sufficient room for the line to be saved. This could cause assertion
failure in forget_input().
Fix spy to reliably avoid spying same sector more than once
spy() stored coordinates of sectors successfully spied in an array.
Since it didn't bother to normalize coordinates, it could spy sectors
near the "seams" of the world more than once. The array was also
wastefully large.
Fix by using a sector bitmap instead, like do_look().
GCFx() and GCFy() pointlessly duplicate sctoff(), and don't work when
the argument is less than -WORLD_X or -WORLD_Y, respectively. With
current code, this happens only for impractically tiny worlds:
emp_setbitmap() writes to bitmap[] out of bounds, and smashes the
heap.
Fix move_ground() to pass coordinate arguments normalized
move_ground() passes curx, cury to various functions that should
probably be called with normalized coordinate arguments. Make sure
curx, cury are normalized.
launch_sat() failed to normalize the satellite's coordinates when it
added a random course deviation. Satellites with screwed up
coordinates were missed by skywatch, and possibly in other places.
Fix recently introduced memory corrupter in path()
path() failed to normalize coordinate argument for deltx() and
delty(). This could subscript map[] and mapbuf[] out of bounds.
Broken in commit 3ca88271.
Ron Koenderink [Sun, 24 Aug 2008 15:09:22 +0000 (09:09 -0600)]
Add server.log and journal.log to nightly build
Turn on journaling using journal.econfig.
Append the server.log and journal.log files
to the nightly build log in the SERVERSTOP step.
Remove the server.log and journal.log in the
SERVERSTART step to ensure only this run's log
events are included in the nightly build log.
Add the construction of a econfig in the
generate step using *.econfig in the patches
directory in a same way code patches are applied.
The mingw.patch was removed and replaced with the
mingw.econfig. The mingw.econfig also required
the addition of exports to the win32.i386.config.
MAPWIDTH(3) used to be one too small, which made rout() clobber the
map row's terminating zero when the map spans the whole world in x.
Since Chainsaw 2, rout() copied rows into an additional buffer to add
a terminating zero. Remove that junk.
MAPWIDTH(x) was (x+2)/2 - 1 too small. Affected were path and route,
which use MAPWIDTH(3) to size their map buffer: they clobber map rows'
terminating zero when the map spans the whole world in x. rout() has
a cheesy work-around for that since Chainsaw 2. path() doesn't, and
duly breaks.
Fix default map size in interactive move, test and transport
Sub-command 'm' calls display_region_map() to display a map. The map
is centered on the current sector by default. It extended one sector
farther to the right and down than to the left and up. Odd, and
inconsistent with the map size used by unit_map() for navigate and
march sub-command 'M'. Fix that.
Rework code dealing with struct range fixing many bugs
Change struct range from exclusive to inclusive upper bounds, for
consistency with struct realmstr and the area syntax. Also fix many
bugs.
real()'s conversion from struct range's exclusive upper bounds to
struct realmstr's inclusive upper bounds could underflow and store -1
in the realms file. Harmless, because its users didn't mind:
list_realm() and nstr_exec_val() convert back to relative coordinates,
and sarg_getrange() is only used by sarg_area(), which happened to
undo the damage. The change to inclusive upper bounds gets rid of the
broken conversion.
xyinrange() incorrectly treated the upper bound as inclusive, unless
the bounds were equal. Impact:
* nxtitem() and nxtitemp() cases NS_AREA and NS_DIST attempted to hack
around xyinrange()'s lossage(!), but screwed up: sectors on the
lower bound of of a range spanning the the whole world were skipped.
This affected all command arguments that support area or distance
syntax for items. In sufficiently small worlds, it could also make
radar miss satellites and ships, sonar miss ships, satellite miss
ships and land units, nuclear detonations miss ships, planes, land
units and nukes, automatic supply miss ship and land unit supply
sources, ships and land units fail to return fire, ships fail to
fire support.
* draw_map() could draw units sitting just right or just below of the
mapped area. No effect, as these parts of the map weren't actually
shown.
xydist_range() produced an inclusive upper bound when it decided that
the range covers everything in that dimension (which it didn't get
quite right either). This could make snxtsct_dist() and
snxtitem_dist() initialize the iterator with an incorrect upper bound.
Similar impact as the xyinrange() / nxtitem() lossage.
border() could print the hundreds line unnecessarily.
snxtsct() and snxtsct_all() screwed up for odd WORLD_Y: they failed to
include (WORLD_Y - 1) / 2 in the y-range. This affected all command
arguments that support "*" syntax for sectors, plus add ... c, power
n, and break.
snxtsct_all() failed to normalize the bounds (presumed harmless).
There were a few correct, but somewhat unclean uses of struct range
with inclusive upper bounds:
* nat_reset() used one internally.
* pathrange() worked with inclusive upper bounds internally, but
corrected to exclusive upper bounds before passing the range out.
* sarg_getrange() worked with inclusive upper bounds. Its only caller
sarg_area() corrected that to exclusive upper bounds.
The change to inclusive upper bounds cleans this up.
unit_map() and xysize_range() had no issues (isn't that amazing?), but
need to be updated for the changed struct range semantics.
Make snxtsct_area() expect correct range width and height
snxtsct_area() computed width and height, overwriting the values
passed in, even though all but two callers passed correct values. The
exceptions were snxtsct() in case NS_ALL, and snxtsct_all(). Change
them to pass correct values, and drop the recomputation from
snxtsct_area(). Simplifies the interface between snxtsct_area() and
its callers.
The correct method to compute indexes into a map buffer for a struct
range is deltx(), delty().
path() used deltax(), deltay() instead, which yield correct results
only for indexes up to half the world size. Pathes spanning larger
areas were screwed up.
sona(), radmap2(), satmap() also used deltax(), deltay(), but only
with arguments where those yield correct results.
draw_map() used xnorm(), ynorm() instead, which is correct, but less
clear and less efficient.
Ron Koenderink [Sun, 17 Aug 2008 21:51:40 +0000 (15:51 -0600)]
Add additional tests to nightly build
The following tests were added:
bridge building
spy
radar (ship, sector)
look
convert
In additon, prepartion for testing research and
hapiness production was added and improvements
for general economy for player 1 were added.
Ron Koenderink [Sun, 17 Aug 2008 21:13:57 +0000 (15:13 -0600)]
Fix timestamps for data files and lost command
Set the ef_timestamp to 100 for the nightly build.
Set the DUMP LOST timestamp to 101 for the nightly build.
This will the suppress the timestamp differences
that occur when testing lost command and will facilitate
testing for dump and xdump commands.
Ron Koenderink [Sat, 9 Aug 2008 21:05:40 +0000 (15:05 -0600)]
Add additional tests to nightly build
The following tests were added:
ship list
ship navigation
assault
tech production
sector lost
commodity monitoring
tech monitoring
education monitoring
Failing a command with code RET_SYN prints help and doesn't charge
BTUs. Failing with code RET_FAIL doesn't print help and charges BTUs.
A couple of command failures were changed or added recently to fail
with RET_SYN, because they're due to invalid player input. Some of
them, however, can happen after the command already did something, so
BTUs must be charged, or else players can deliberately fail the
command to save BTUs:
* Commit 9eda5f87 adds RET_SYN failures when getting player input
fails for:
- arm third argument
- deliver fourth argument
- fire third argument
- lmine second argument
- order d fourth argument
- range second argument
- sail second argument
- tend third argument
- designate second argument
- morale second argument
- set third argument
- tend fourth argument
* Commit d000bf92 likewise (with a bogus commit message) for bdes
second argument.
* Commit 9f4ce71a likewise for ltend third and fourth argument.
* Commit 9031b03b changes failure code from RET_FAIL when getting
player input fails for threshold third argument. It adds RET_SYN
failure when the argument is bad. Some bad arguments already failed
that way before.
* Commit a7cf69af changes it from RET_FAIL when designate second
argument is bad.
Change them all to fail with RET_FAIL.
Many other places have the same bug, but those are left for another
day.
Improve work()'s message when sector doesn't need construction
The cheesy test for repeated messages broke down when working on more
than one sector. Reword the message so that repetition is fine, and
drop the test.
Don't permit convert to spend more money than available
Complicated by the fact that conv() ran the conversion code twice,
first for adding up the cost for chkmoney(), then for actually
converting. chkmoney() asks the player to confirm when he's about to
spend more than half his cash. Get rid of that, not worth the
complexity. This merges do_conv() back into conv().
Don't permit demobilize to spend more money than available
Complicated by the fact that demo() ran the demobilization code twice,
first for adding up the cost for chkmoney(), then for actually
demobilizing. chkmoney() asks the player to confirm when he's about
to spend more than half his cash. Get rid of that, not worth the
complexity. This merges do_demo() back into demo().
It also removes the command's virtually undocumented fourth argument.
Update player_coms[] accordingly. While there, make it require money;
it won't do anything useful without money anyway.
RET_SYS was used for commands failing due to internal or environmental
errors, but not really systematically. The difference to RET_FAIL is
how dispatch() treats them: RET_SYS got logged, and cost no BTUs.
More specific logging is possible at the point of failure than in
dispatch(). Make sure that's done for all failures that used to
return RET_SYS.
The change in BTU charging affects commands consider, offer, repay,
trade failing due to internal errors. It also affects deity commands
reload and turn (irrelevant because deities get unlimited BTUs), and
commands apropos, info and motd (irrelevant because they cost no
BTUs).
projects / empserver / log