oweals/minetest.git
8 years agoAdd base64 encoding and decoding to the lua api. (#3919)
red-001 [Sat, 28 May 2016 03:37:28 +0000 (04:37 +0100)]
Add base64 encoding and decoding to the lua api. (#3919)

8 years agoDisallow external linkage for default Cavegen NoiseParams
kwolekr [Fri, 27 May 2016 03:23:05 +0000 (23:23 -0400)]
Disallow external linkage for default Cavegen NoiseParams

8 years agoDungeongen: Remove dependency on Mapgen
kwolekr [Sun, 22 May 2016 20:50:43 +0000 (16:50 -0400)]
Dungeongen: Remove dependency on Mapgen

8 years agoMapgen: Combine dungeon generation code
kwolekr [Sun, 22 May 2016 20:27:31 +0000 (16:27 -0400)]
Mapgen: Combine dungeon generation code

8 years agoMapgen: Deduplicate common constructor code
kwolekr [Sun, 22 May 2016 06:17:19 +0000 (02:17 -0400)]
Mapgen: Deduplicate common constructor code

8 years agoMapgen: Remove calculateNoise from most mapgens
kwolekr [Fri, 20 May 2016 07:37:31 +0000 (03:37 -0400)]
Mapgen: Remove calculateNoise from most mapgens

This commit moves noise calculation to the functions where the noise is
actually required, increasing the separation of concerns and level of
interdependency for each mapgen method.  Valleys Mapgen is left unmodified.

8 years agoMapgen V7: Remove now-unused ridge heightmap
kwolekr [Thu, 19 May 2016 18:05:51 +0000 (14:05 -0400)]
Mapgen V7: Remove now-unused ridge heightmap

8 years agoCavegen: Move V5-style caves to CavesNoiseIntersection
kwolekr [Thu, 19 May 2016 07:40:22 +0000 (03:40 -0400)]
Cavegen: Move V5-style caves to CavesNoiseIntersection

8 years agoCavegen: Minor misc. fixes
kwolekr [Sun, 15 May 2016 02:46:50 +0000 (22:46 -0400)]
Cavegen: Minor misc. fixes

- Convert instances of numeric literal doubles to floats
- Move dswitchint to a local variable now that being a member is unnecessary
- Improve const correctness
- Indentation fixes

8 years agoFix undefined evaluation order when constructing random vectors
kwolekr [Wed, 11 May 2016 07:47:45 +0000 (03:47 -0400)]
Fix undefined evaluation order when constructing random vectors

MSVC and GCC evaluate parameters in right-to-left order, whereas Clang
evaluates in left-to-right order, and of course, an optimization could
leave the order of evaluation completely indeterminate.

This commit fixes all instances of the error by explicitly assigning the
results of expressions that use PseudoRandom::next() or range() to their
respective vector components.

The right-to-left evaluation behavior is preserved since Clang is much less
commonly used to compile Minetest than GCC and MSVC combined, and would
therefore cause the least harm.

8 years agoCavegen: Re-add small caves to CavesRandomWalk
kwolekr [Wed, 11 May 2016 04:00:05 +0000 (00:00 -0400)]
Cavegen: Re-add small caves to CavesRandomWalk

8 years agoCavegen: Remove CavesRandomWalk dependency on Mapgen
kwolekr [Wed, 11 May 2016 02:56:03 +0000 (22:56 -0400)]
Cavegen: Remove CavesRandomWalk dependency on Mapgen

8 years agoCavegen: Merge instances of repetitive surface level-finding code
kwolekr [Tue, 10 May 2016 17:35:39 +0000 (13:35 -0400)]
Cavegen: Merge instances of repetitive surface level-finding code

8 years agoCavegen: Remove CavesV6 dependency on Mapgen
kwolekr [Tue, 10 May 2016 17:21:13 +0000 (13:21 -0400)]
Cavegen: Remove CavesV6 dependency on Mapgen

8 years agoCavegen: Rename CaveV6 to CavesV6
kwolekr [Tue, 10 May 2016 04:10:59 +0000 (00:10 -0400)]
Cavegen: Rename CaveV6 to CavesV6

- Add comment explaining why it exists
- Remove unused 'flooded' variable
- Rename shadowed variable
- Fix some code style

8 years agoCavegen: Merge CaveV5 and CaveV7 into CavesRandomWalk
kwolekr [Tue, 10 May 2016 04:01:37 +0000 (00:01 -0400)]
Cavegen: Merge CaveV5 and CaveV7 into CavesRandomWalk

8 years agoFix MgStoneType and BiomeType enum names
kwolekr [Mon, 2 May 2016 06:45:59 +0000 (02:45 -0400)]
Fix MgStoneType and BiomeType enum names

8 years agoMapgen: Combine generateBiomes, dustTopNodes, and generateCaves
kwolekr [Mon, 2 May 2016 06:24:57 +0000 (02:24 -0400)]
Mapgen: Combine generateBiomes, dustTopNodes, and generateCaves

This commit condenses the above methods into a single implementation used by
V7, V5, Flat, Fractal, and Valleys mapgens and introduces MapgenBasic.

8 years agoMove biome calculation to BiomeGen
kwolekr [Thu, 28 Apr 2016 07:43:09 +0000 (03:43 -0400)]
Move biome calculation to BiomeGen

BiomeGen defines an interface that, given a set of BiomeParams, computes biomes
for a given area using the algorithm implemented by that specific BiomeGen.
This abstracts away the old system where each mapgen supplied the noises
required for biome generation.

8 years agoTell irrlicht if we handle a key or not.
est31 [Sat, 14 May 2016 14:25:57 +0000 (16:25 +0200)]
Tell irrlicht if we handle a key or not.

We can remove the function in MtNativeActivity now
as it serves precisely that purpose: to tell irrlicht
that we handled the esc key.

TODO for later:
 * Perhaps try to find a more performant container than KeyList

8 years agoFix rollback.txt migration
ShadowNinja [Fri, 12 Feb 2016 23:39:03 +0000 (18:39 -0500)]
Fix rollback.txt migration

Broken by b1965ac20922e3722392114bd63a22b403dcbe98.
This also prepares the begin and commit statements only once.

8 years agoFix irrlicht version checking macro for tooltip_height calculation
Craig Robbins [Mon, 23 May 2016 14:27:11 +0000 (00:27 +1000)]
Fix irrlicht version checking macro for tooltip_height calculation

8 years agoMove updateTextures and fillTileAttribs to ContentFeatures
Ekdohibs [Sun, 22 May 2016 18:33:06 +0000 (20:33 +0200)]
Move updateTextures and fillTileAttribs to ContentFeatures

8 years agoTolerate packet reordering in the early init process 4140/head
est31 [Wed, 18 May 2016 04:18:08 +0000 (06:18 +0200)]
Tolerate packet reordering in the early init process

Fixes a bug where packet reordering made the server give the
client two peer ids instead of one. This in turn confused
reliable packet sending and made connecting to the server fail.

The client usually sends three packets at init: one "dummy"
packet consisting of two 0 bytes, and the init packet as well as
its legacy counterpart. The last one can be turned off since commit
af30183124d40a969040d7de4b3a487feec466e4, but this is of lower
relevance for the bug. The relevant part here is that network
packet reorder (which is a normal occurence) can make the packets
reach the server in different order.

If reorder puts the dummy packet further behind, the following
would happen before the patch:

1. The server will get one of the init packets on channel 1 and
   assign the client a peer id, as the packet will have zero as
   peer id.

2. The server sends a CONTROLTYPE_SET_PEER_ID packet to inform
   the client of the peer id.

3. The next packet from the client will contain the peer id set by
   the server.

4. The server sets the m_has_sent_with_id member for the client's
   peer structure to true.

5. Now the dummy packet arrives. It has a peer id of zero, therefore
   the server searches whether it already has a peer id for the
   address the packet was sent from. The search fails because
   m_has_sent_with_id was set to true and the server only searched
   for peers with m_has_sent_with_id set to false.

6. In a working setup, the server would assign the dummy packet to
   the correct peer id. However the server instead now assigns a
   second peer id and peer structure to the peer, and assign the
   packet to that new peer.

7. In order to inform the peer of its peer id, the server sends a
   CONTROLTYPE_SET_PEER_ID command packet, reliably, to the peer.
   This packet uses the new peer id.

8. The client sends an ack to that packet, not with the new peer id
   but with the peer id sent in 2.

9. This packet reaches the server, but it drops the ACK as the peer
   id does not map to any un-ACK-ed packets with that seqnum. The
   same time, the server still waits for an ACK with the new peer
   id, which of course won't come. This causes the server to
   periodically re-try sending that packet, and the client ACKing it
   each time.

Steps 7-9 cause annoyances and erroneous output, but don't cause
the connection failure itself.
The actual mistake that causes the connection failure happens in 6:
The server does not assign the dummy packet to the correct peer, but
to a newly created one.
Therefore, all further packets sent by the client on channel 0 are
now buffered by the server as it waits for the dummy packet to reach
the peer, which of course doesn't happen as the server assigned
that packet to the second peer it created for the client.
This makes the connection code indefinitely buffer the
TOSERVER_CLIENT_READY packet, not passing it to higher level code,
which stalls the continuation of the further init process
indefinitely and causes the actual bug.

Maybe this can be caused by reordered init packets as well, the only
studied case was where network has reliably reordered the dummy
packet to get sent after the init packets.

The patch fixes the bug by not ignoring peers where
m_has_sent_with_id has been set anymore. The other changes of the
patch are just cleanups of unused methods and fields and additional
explanatory comments.

One could think of alternate ways to fix the bug:

* The client could simply take the new peer id and continue
  communicating with that. This is however worse than the fix as
  it requires the peer id set command to be sent reliably (which
  currently happens, but it cant be changed anymore). Also, such a
  change would require both server and client to be patched in order
  for the bug to be fixed, as right now the client ignores peer id
  set commands after the peer id is different from
  PEER_ID_INEXISTENT and the server requires modification too to
  change the peer id internally.
  And, most importantly, right now we guarantee higher level server
  code that the peer id for a certain peer does not change. This
  guarantee would have to be broken, and it would require much
  larger changes to the server than this patch means.

* One could stop sending the dummy packet. One may be unsure whether
  this is a good idea, as the meaning of the dummy packet is not
  known (it might be there for something important), and as it is
  possible that the init packets may cause this problem as well
  (although it may be possible too that they can't cause this).

Thanks to @auouymous who had originally reported this bug and who
has helped patiently in finding its cause.

8 years agoFix a m_camera not used warning fix pointed by clang 4154/head
Loic Blot [Sun, 22 May 2016 11:31:41 +0000 (13:31 +0200)]
Fix a m_camera not used warning fix pointed by clang

8 years agoImplement a PostgreSQL backend 4124/head
Loic Blot [Sat, 14 May 2016 09:00:42 +0000 (11:00 +0200)]
Implement a PostgreSQL backend

8 years agoGitignore: ignore idea and ninja files
HybridDog [Fri, 20 May 2016 20:08:07 +0000 (22:08 +0200)]
Gitignore: ignore idea and ninja files

8 years agoItem entities: Don't show description as infotext
paramat [Mon, 16 May 2016 23:22:27 +0000 (00:22 +0100)]
Item entities: Don't show description as infotext

Partially reverts #3547
Infotext remains optional for objects, empty by default

8 years agoFix tooltip height for versions of irrlicht < 1.8.2
Craig Robbins [Fri, 20 May 2016 11:18:59 +0000 (21:18 +1000)]
Fix tooltip height for versions of irrlicht < 1.8.2

Version 1.8.2 of irrlicht changed the way that IGUIStaticText::getTextHeight() works and since that release properly deals with newlines.

From irrlicht changes.txt for 1.8.2, "IGUIStaticText::getTextHeight returns now the correct height for texts with newlines even WordWrap is not set."

8 years agoImplement DatabaseException for databases 4136/head
nerzhul [Tue, 17 May 2016 07:59:02 +0000 (09:59 +0200)]
Implement DatabaseException for databases

8 years agoAdd on_punchnode callback
Wayward One [Sat, 30 Apr 2016 14:31:11 +0000 (10:31 -0400)]
Add on_punchnode callback

8 years agoDB::loadBlock copy removal & DB backend cleanup 4125/head
Loic Blot [Sat, 14 May 2016 10:23:15 +0000 (12:23 +0200)]
DB::loadBlock copy removal & DB backend cleanup
* Remove the copy from db::loadBlock by using a pointer to the destination
* cleanup db backend, the child backend doesn't have to set their functions as virtual

8 years agoReally fix ncurses lookup on Arch Linux
sfan5 [Mon, 16 May 2016 10:42:51 +0000 (12:42 +0200)]
Really fix ncurses lookup on Arch Linux

Commit 27ee8d8943080a5dd735c9faa47c726604bafdff forgot to add the paths
without ncursesw/ to the find_path() call

8 years agoAndroid: enable parallelism for main target too
est31 [Thu, 12 May 2016 22:14:24 +0000 (00:14 +0200)]
Android: enable parallelism for main target too

This adds to the changes that commit

98d16e0d9a945f5f48462c05f26ae4bde2db5731 "Android: Tell make about sub-makes to speed up build"

did, and enables parallel builds for minetest
itself as well.

8 years agoDocumentation: Remove incorrect and excessive mapgen flags text
paramat [Thu, 12 May 2016 03:39:05 +0000 (04:39 +0100)]
Documentation: Remove incorrect and excessive mapgen flags text

8 years agoContinue with 0.4.14-dev
sfan5 [Sun, 15 May 2016 12:49:15 +0000 (14:49 +0200)]
Continue with 0.4.14-dev

8 years agoBump version to 0.4.14 0.4.14
sfan5 [Sun, 15 May 2016 12:49:15 +0000 (14:49 +0200)]
Bump version to 0.4.14

8 years agoAndroid: Re-add button to remove singleplayer world
est31 [Sat, 14 May 2016 20:12:38 +0000 (22:12 +0200)]
Android: Re-add button to remove singleplayer world

Fixes #4120.

8 years agoFix android build by fixing patch line endings
est31 [Sat, 14 May 2016 14:46:46 +0000 (16:46 +0200)]
Fix android build by fixing patch line endings

8 years agoFix locked hardware buttons on Android
Maksim Gamarnik [Sat, 14 May 2016 13:35:56 +0000 (23:35 +1000)]
Fix locked hardware buttons on Android

Fixes #2122
Fixes #1454

Addendum (est31)
According from its docs in android_native_app_glue.h (from the NDK), the
onInputEvent should "Return 1 if you have handled the event, 0 for any
default dispatching". Before, we always returned 1, meaning we blocked
all hardware keys to be given to the OS.
This broke the volume keys and has caused #2122 and #1454.

Although it bases on lots of guesswork, it can probably safely be said that
CGUIEnvironment::postEventFromUser returns true if the event was handled,
and false if not. Therefore, set the status variable depending on what
postEventFromUser returned.

8 years agoMainmenu: don't show change keys dialog on android
est31 [Fri, 13 May 2016 00:22:31 +0000 (02:22 +0200)]
Mainmenu: don't show change keys dialog on android

The change keys dialog can't be left. It doesn't make
much sense to show it on Android in the first place,
therefore disable it, just like commit

aed70cb0b652d6cb2272e7b94cd56671b3df6239 'Disable sound and key binding settings in "pause" menu on android'

has disabled it for the esc menu.

Fixes #4115.

8 years agoFix #4111 ("Provided world path doesn't exist" if choosing singleplayer on first...
Rui [Thu, 12 May 2016 15:38:29 +0000 (01:38 +1000)]
Fix #4111 ("Provided world path doesn't exist" if choosing singleplayer on first run")

Bug and whitespace error fixed (Zeno)

8 years agoAndroid: download deps using https
est31 [Thu, 12 May 2016 00:11:27 +0000 (02:11 +0200)]
Android: download deps using https

Its more secure, and some pages even redirect to the https version.

8 years agoAndroid: update openssl to 1.0.2h
est31 [Wed, 11 May 2016 22:52:29 +0000 (00:52 +0200)]
Android: update openssl to 1.0.2h

8 years agoFix typo in the info printed by --version
Craig Robbins [Wed, 11 May 2016 13:20:20 +0000 (23:20 +1000)]
Fix typo in the info printed by --version

8 years agoAdd platform name to --version information
Craig Robbins [Wed, 11 May 2016 07:15:41 +0000 (17:15 +1000)]
Add platform name to --version information

8 years agoMapgen V6: Synchronize spflags with defaultsettings
kwolekr [Wed, 11 May 2016 02:06:03 +0000 (22:06 -0400)]
Mapgen V6: Synchronize spflags with defaultsettings

This fixes an issue where trees are omitted from Mapgen V6 maps on
configurations that explicitly defined the mgv6_spflags setting.

8 years agoREADME: Update Fedora dependencies list
Johnny [Tue, 10 May 2016 21:28:45 +0000 (14:28 -0700)]
README: Update Fedora dependencies list

Edited packages to reflect correct packages

8 years agoTranslated using Weblate (Spanish)
Diego Martínez [Fri, 6 May 2016 01:34:11 +0000 (03:34 +0200)]
Translated using Weblate (Spanish)

Currently translated at 56.8% (504 of 887 strings)

This is a merger of two commits.

8 years agoTranslated using Weblate (Spanish)
Lucas Montenegro [Tue, 10 May 2016 15:32:35 +0000 (17:32 +0200)]
Translated using Weblate (Spanish)

Currently translated at 51.5% (457 of 887 strings)

This is a merger of two commits.

8 years agoTranslated using Weblate (Spanish)
ShadowNinja [Tue, 10 May 2016 15:34:28 +0000 (17:34 +0200)]
Translated using Weblate (Spanish)

Currently translated at 51.5% (457 of 887 strings)

8 years agoTranslated using Weblate (Ukrainian)
Fixer [Mon, 9 May 2016 14:22:10 +0000 (16:22 +0200)]
Translated using Weblate (Ukrainian)

Currently translated at 38.4% (341 of 887 strings)

This is a merger of two commits.

8 years agoTranslated using Weblate (Lojban)
senpi [Sun, 8 May 2016 15:39:07 +0000 (17:39 +0200)]
Translated using Weblate (Lojban)

Currently translated at 10.5% (94 of 887 strings)

8 years agoTranslated using Weblate (Italian)
Emon Omen [Sat, 7 May 2016 18:15:36 +0000 (20:15 +0200)]
Translated using Weblate (Italian)

Currently translated at 100.0% (887 of 887 strings)

This is a merger of three commits.

8 years agoTranslated using Weblate (Japanese)
BreadW [Sat, 7 May 2016 14:10:10 +0000 (16:10 +0200)]
Translated using Weblate (Japanese)

Currently translated at 51.5% (457 of 887 strings)

8 years agoTranslated using Weblate (French)
Yvan YR [Sat, 7 May 2016 09:19:25 +0000 (11:19 +0200)]
Translated using Weblate (French)

Currently translated at 90.7% (805 of 887 strings)

8 years agoTranslated using Weblate (Japanese)
BreadW [Fri, 6 May 2016 22:18:04 +0000 (00:18 +0200)]
Translated using Weblate (Japanese)

Currently translated at 50.6% (449 of 887 strings)

8 years agoTranslated using Weblate (Hungarian)
Kisbenedek Márton [Sat, 7 May 2016 01:17:37 +0000 (03:17 +0200)]
Translated using Weblate (Hungarian)

Currently translated at 75.7% (672 of 887 strings)

8 years agoTranslated using Weblate (Danish)
Thomas Wagner Nielsen [Thu, 5 May 2016 20:50:35 +0000 (22:50 +0200)]
Translated using Weblate (Danish)

Currently translated at 27.6% (245 of 887 strings)

This is a merger of two commits.

8 years agoTranslated using Weblate (Chinese (China))
Claybiokiller [Thu, 5 May 2016 17:09:42 +0000 (19:09 +0200)]
Translated using Weblate (Chinese (China))

Currently translated at 73.9% (656 of 887 strings)

8 years agoTranslated using Weblate (German)
Wuzzy [Thu, 5 May 2016 23:16:37 +0000 (01:16 +0200)]
Translated using Weblate (German)

Currently translated at 100.0% (887 of 887 strings)

This is a merger of three commits.

8 years agoTranslated using Weblate (Dutch)
Auke Kok [Thu, 5 May 2016 20:33:07 +0000 (22:33 +0200)]
Translated using Weblate (Dutch)

Currently translated at 89.6% (795 of 887 strings)

This is a merger of three commits.

8 years agoTranslated using Weblate (Dutch)
Tobyplowy [Thu, 5 May 2016 20:33:12 +0000 (22:33 +0200)]
Translated using Weblate (Dutch)

Currently translated at 85.6% (760 of 887 strings)

8 years agoLua_api.txt: Fix documentation for facedir rotation
paramat [Mon, 9 May 2016 22:44:15 +0000 (23:44 +0100)]
Lua_api.txt: Fix documentation for facedir rotation

8 years agoAdd [resize texture modifier Resizes the texture to the given dimensions.
SmallJoker [Tue, 3 May 2016 18:14:01 +0000 (20:14 +0200)]
Add [resize texture modifier Resizes the texture to the given dimensions.

8 years agoUpdate credits tab 3907/head
est31 [Thu, 24 Mar 2016 16:17:37 +0000 (17:17 +0100)]
Update credits tab

8 years agoDefaultsettings/Android: Increase 'max block generate distance' to 3
paramat [Sun, 8 May 2016 00:03:56 +0000 (01:03 +0100)]
Defaultsettings/Android: Increase 'max block generate distance' to 3

8 years agoFixes #4098 ce8a9ed didn't quite go far enough, and left this bug in
gregorycu [Sun, 8 May 2016 09:02:11 +0000 (19:02 +1000)]
Fixes #4098 ce8a9ed didn't quite go far enough, and left this bug in

8 years agoChat: Keep scroll position constant in ChatBuffer::deleteOldest()
Kahrl [Tue, 23 Apr 2013 21:17:33 +0000 (23:17 +0200)]
Chat: Keep scroll position constant in ChatBuffer::deleteOldest()

8 years agoOptimize default settings for Android build
Maksim Gamarnik [Sun, 10 Apr 2016 00:12:10 +0000 (03:12 +0300)]
Optimize default settings for Android build

8 years agoRun unescape_enriched *after* unescape_string.
Ekdohibs [Thu, 5 May 2016 17:08:45 +0000 (19:08 +0200)]
Run unescape_enriched *after* unescape_string.

Doing it the other way round was a mistake, since it breaks
minetest.formspec_escape with escape sequences that contain
special characters.

8 years agoMake dropdowns show the string that was their argument.
Ekdohibs [Thu, 5 May 2016 16:50:02 +0000 (18:50 +0200)]
Make dropdowns show the string that was their argument.

This makes it work even if it contains escape sequences,
which didn't work before.

8 years agoAndroid: Increase player_stepheight for thicker snow nodebox
Maksim Gamarnik [Tue, 3 May 2016 00:24:36 +0000 (03:24 +0300)]
Android: Increase player_stepheight for thicker snow nodebox

8 years agoLua_api.txt: Add warnings of l-system lighting bug
paramat [Mon, 2 May 2016 20:22:06 +0000 (21:22 +0100)]
Lua_api.txt: Add warnings of l-system lighting bug

8 years agoRun updatepo.sh
est31 [Thu, 5 May 2016 14:14:09 +0000 (16:14 +0200)]
Run updatepo.sh

8 years agoUpdate settings translation file and minetest.conf.example
est31 [Thu, 5 May 2016 14:12:58 +0000 (16:12 +0200)]
Update settings translation file and minetest.conf.example

8 years agoFix holding down F10 (open console) causing GUI to freeze
Craig Robbins [Wed, 4 May 2016 11:21:13 +0000 (21:21 +1000)]
Fix holding down F10 (open console) causing GUI to freeze

8 years agoSettings_translation_file: Update mapgen with cave width parameter
paramat [Mon, 2 May 2016 22:16:28 +0000 (23:16 +0100)]
Settings_translation_file: Update mapgen with cave width parameter

8 years agoFix Windows build
Craig Robbins [Mon, 2 May 2016 05:01:17 +0000 (15:01 +1000)]
Fix Windows build

Fixes the issue introduced by c1a0ebb (Fix use of uninitialised variable
in class Event) causing Windows builds to fail

8 years agoFileSelectMenu: Fix formspec parsing broken by Irrlicht file-chooser
Kahrl [Thu, 28 Apr 2016 18:27:59 +0000 (20:27 +0200)]
FileSelectMenu: Fix formspec parsing broken by Irrlicht file-chooser

8 years agofind_path: consider walkable instead of CONTENT_AIR 4037/head
Auke Kok [Tue, 26 Apr 2016 00:14:57 +0000 (17:14 -0700)]
find_path: consider walkable instead of CONTENT_AIR

The path finding code works fairly well except that it considers
anythin not CONTENT_AIR to be "above the surface". This results in
paths that are unwalkable for entities since e.g. plants are not
walkable. The path would force them to jump on top of grass plants,
etc..

The obvious solution is not to use CONTENT_AIR as a criteria, but
instead distinguish between walkable and non-walkable nodes. This
results in paths that properly walk through grass nodes.

This was extensively tested by a flock of electric sheep.

Note that for underwater purposes this changes the behaviour from
"the surface is walkable" to "ignore water entirely" making the
path go across the water bottom, and pathing fail likely from the
water surface. This is intentional.

8 years agoPathfinder: improve GridNode storage
est31 [Fri, 1 Apr 2016 01:13:24 +0000 (03:13 +0200)]
Pathfinder: improve GridNode storage

Before, the GridNodes were stored in vector<vector<vector<T>>>,
and initialized in advance. Putting three vectors inside each other
puts lots of unneccessary stress onto the allocator, costs more memory,
and has worse cache locality than a flat vector<T>.

For larger search distances, an the array getting initialized means
essentially O(distance^3) complexity in both time and memory,
which makes the current path search a joke. In order to really
profit from the dijkstra/A* algorithms, other data structures
need to be used for larger distances.

For shorter distances, a map based GridNode storage may be slow as
it requires lots of levels of indirection, which is bad for things like
cache locality, and an array based storage may be faster.

This commit does:

1. remove the vector<vector<vector<T>>> based GridNodes storage that
is allocated and initialized in advance and for the whole
possible area.

2. Add a vector<T> based GridNodes storage that is allocated and
initialized in advance for the whole possible area.

3. Add a map<P,T> based GridNodes storage whose elements are
allocated and initialized, when the path search code
demands it.

4. Add code to decide between approach 2 and 3,
based on the length of the path.

5. Remove the unused "surfaces" member of the PathGridnode class.
Setting this isn't as easy anymore for the
map based GridNodes storage.

8 years agoPathfinder: use core::aabbox3d instead of own type
est31 [Fri, 1 Apr 2016 00:43:28 +0000 (02:43 +0200)]
Pathfinder: use core::aabbox3d instead of own type

There is no need to reinvent the wheel here, we have
great classes from irrlicht.

8 years agoPathfinder: Fix style
est31 [Thu, 31 Mar 2016 23:52:17 +0000 (01:52 +0200)]
Pathfinder: Fix style

* Fix naming style for methods and classes:
Use camelCase for methods and PascalCase for classes as
code style demands it. And use sneak_case for methods that
are not member of a class.
* Replace "* " with " *" for Pointers
* Same for references
* Put function body opening braces on new line
* Other misc minor non functional style improvements

8 years agoMove pathfinder classes to cpp file
est31 [Thu, 31 Mar 2016 23:20:34 +0000 (01:20 +0200)]
Move pathfinder classes to cpp file

There is no need to put them into the header, they are solely used
inside the pathfinder.

Another advantage of this change is that only the pathfinder.cpp has
to be compiled if PATHFINDER_DEBUG gets defined or undefined, not
all files including the .h.

This commit moves the pathfinder classes to the cpp file without
modifications.
Also, the PATHFINDER_DEBUG macro gets moved to the cpp file and
the PATHFINDER_CALC_TIME macro gets moved to a plce where it
actually does work.

8 years agoFix use of uninitialised variable in class Event
Craig Robbins [Sun, 1 May 2016 06:47:08 +0000 (16:47 +1000)]
Fix use of uninitialised variable in class Event

8 years agoUse MoveFileEx to rename files on Windows (not rename)
gregorycu [Sun, 1 May 2016 07:27:29 +0000 (17:27 +1000)]
Use MoveFileEx to rename files on Windows (not rename)

8 years agoMainmenu: Remove space under mod list
Rui [Sat, 30 Apr 2016 08:09:59 +0000 (17:09 +0900)]
Mainmenu: Remove space under mod list

8 years agoTranslated using Weblate (Japanese)
Rui [Wed, 30 Mar 2016 11:09:05 +0000 (13:09 +0200)]
Translated using Weblate (Japanese)

Currently translated at 51.7% (448 of 865 strings)

This is a merger of two commits.

8 years agoTranslated using Weblate (Italian)
Emon Omen [Fri, 29 Apr 2016 19:48:10 +0000 (21:48 +0200)]
Translated using Weblate (Italian)

Currently translated at 100.0% (865 of 865 strings)

This is a merger of two commits.

8 years agoTranslated using Weblate (Chinese (China))
Claybiokiller [Sun, 24 Apr 2016 13:15:50 +0000 (15:15 +0200)]
Translated using Weblate (Chinese (China))

Currently translated at 75.4% (653 of 865 strings)

8 years agoTranslated using Weblate (Catalan)
Joan Ciprià Moreno Teodoro [Sat, 16 Apr 2016 09:26:36 +0000 (11:26 +0200)]
Translated using Weblate (Catalan)

Currently translated at 46.5% (403 of 865 strings)

This is a merger of two commits.

8 years agoTranslated using Weblate (Danish)
Thomas Wagner Nielsen [Sun, 17 Apr 2016 06:53:06 +0000 (08:53 +0200)]
Translated using Weblate (Danish)

Currently translated at 28.7% (249 of 865 strings)

8 years agoTranslated using Weblate (Indonesian)
Muhammad Rifqi Priyo Susanto [Tue, 12 Apr 2016 16:24:43 +0000 (18:24 +0200)]
Translated using Weblate (Indonesian)

Currently translated at 57.4% (497 of 865 strings)

This is a merger of three commits.

8 years agoTranslated using Weblate (Estonian)
Jan Harald [Sat, 9 Apr 2016 11:53:33 +0000 (13:53 +0200)]
Translated using Weblate (Estonian)

Currently translated at 21.8% (189 of 865 strings)

8 years agoTranslated using Weblate (Portuguese)
Fernando Reis [Fri, 25 Mar 2016 12:26:39 +0000 (13:26 +0100)]
Translated using Weblate (Portuguese)

Currently translated at 67.5% (584 of 865 strings)

This is a merger of 5 commits.

8 years agoTranslated using Weblate (German)
Wuzzy [Sun, 27 Mar 2016 01:07:17 +0000 (03:07 +0200)]
Translated using Weblate (German)

Currently translated at 100.0% (865 of 865 strings)

This is a merger of two commits.

8 years agoTranslated using Weblate (Chinese (China))
YFdyh000 [Tue, 29 Mar 2016 15:59:08 +0000 (17:59 +0200)]
Translated using Weblate (Chinese (China))

Currently translated at 75.3% (652 of 865 strings)

This is a merger of 3 commits.

8 years agoTranslated using Weblate (Russian)
Stas Kies [Sun, 27 Mar 2016 23:49:35 +0000 (01:49 +0200)]
Translated using Weblate (Russian)

Currently translated at 59.3% (513 of 865 strings)

This is a merger of 5 commits.

8 years agoTranslated using Weblate (Russian)
Pavel Sokolov [Sun, 27 Mar 2016 23:50:02 +0000 (01:50 +0200)]
Translated using Weblate (Russian)

Currently translated at 58.9% (510 of 865 strings)

This is a merger of two commits.