| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
I had broken this in 264be2108c51837fa92085f6c839e66aebbcfc9e by no
longer reading out the property.
Made some further simplifications and two more fixes:
* When the original image's size was not 800x600, it wouldn't get scaled
as expected due to the integer division being performed before the
multiplication.
* I had changed the default resolution to 1280x720, while mKeepRatio
expects to scale the original image as if it was displayed on an
800x600 screen. So now it's hardcoded on 800x600 rather than using
the defaultScreenWidth/Height variables.
* Removed the confusing x/y parameters from AmbientLayer::draw, which
were actually set to the graphics width/height.
* Changed SDLGraphics::drawRescaledImagePattern to not repeatedly
set the source origin.
|
|
Rewrote the tile animation loading code based on XML tags, replacing
the code that loaded tile animations from tile properties.
Also made a number of code simplifications and optimizations:
* Replaced a number of pointer members with value members.
* Pass around Animation and TileAnimation by value, using std::move to
avoid allocating copies.
* push -> emplace
* push_front -> emplace_front
* push_back -> emplace_back
* Use range-based for loops
* Use std::vector instead of std::list for storing affected tiles
(less fragmentation)
* Avoid string copies and allocations while parsing CSV layer data.
* Replaced xmlNodeGetContent with directly accessing 'content'.
|
|
The affected tiles were only tracked for base64-encoded layer data.
|
|
* Use final for all message handlers, Client, LocalPlayer,
Being::getType, Being::setPosition and Being::setMap.
(avoids some warnings about virtual dispatch in constructors)
* Use auto in more places
* Use emplace_back instead of push_back in some places
* Use default member initializers
* Less else after return
* Removed superfluous .c_str()
* Removed type aliases that are only used once
* Removed more unused includes
|
|
This change fixes hair style to take into account "race", which makes
the faces visible again. Hair colors should also be fixed now, with
partial support for itemcolors.xml added.
The Mana client now also supports per-character gender, and it now hides
the hair style and color buttons on character creation, when there are
none to choose from.
Closes #43
|
|
This only adds support for the NPC_CLOSE_DIALOG and
NPC_CLEAR_DIALOG sub-commands.
Closes #47
|
|
* Fixed duplication of player name in chat
Closes #49
|
|
Also made it exit with error when there is an unknown option or missing
argument.
|
|
|
|
Usually this would be guessed correctly by the port, but now it is also
possible to just specify the server type and the port will be derived
from there, unless a default port is given in the branding file.
Closes #56
|
|
To match support for this in M+, since TMW uses this functionality on
two of its NPCs.
|
|
The new cursor types are not used yet for now.
|
|
* Don't needlessly store or return raw pointers in BeingInfo
* Less copying, more moving
* Less else after return
* Make AddDEF a template instead of a macro
* Removed some unused includes
* Use range-based for loops
|
|
The Mana World currently likes to just update its "TMW.zip" file,
whereas updates were always given unique names in the past. With this
change, the client checks the Adler32 checksum to know when it should
re-download an update file.
This matches the behavior of ManaPlus commit
96150f1aeacf55d311c41ffe12d9e754b1cda001.
|
|
Using sprintf triggers a warning on macOS. The buffer should be large
enough, but a little extra security doesn't hurt.
|
|
The configuration and setup UI were adjusted to the new options.
This also fixes issues in applying new video settings.
Default resolution was changed from 800x600 to 1280x720. VSync is
enabled by default while FPS limit was disabled.
Display aspect ratio for the resolution options.
I had to work around some macOS issues:
* Don't change window size when it appears to be "maximized", since it
just changes the rendering area while leaving the window maximized.
* Unset fullscreen display mode temporarily to allow changing resolutions,
otherwise the rendering area no longer matches the screen and mouse
input is also off.
* Removed SDL_WINDOW_ALLOW_HIGHDPI for now because it causes issues on
macOS, since we're not actually handling the scaling factor.
A Video class and an SDLGraphics subclass were split off from Graphics.
This setup has Less duplication and leaves the OpenGLGraphics and
SDLGraphics better separated.
Fixes #57
Fixes #58
|
|
* Use default member initializers
* Use range-based loops
* Don't use 'else' after 'return'
* Removed some unused includes
* Construct empty strings with std::string() instead of ""
* Clear strings with .clear() instead of assigning ""
* Check whether strings are empty with .empty() instead of comparing to ""
* Removed redundant initializations
|
|
Unfortunately, since we're making a difference between config and data,
and also want to know where to write screenshots, we can't get rid of
PHYSFS_getUserDir entirely. We'd need a replacement to get rid of the
deprecation warnings.
Also removed copying of old config file at "/.tmw/config.xml".
|
|
Since Guichan is currently disabled on Homebrew, I've opted for adding
it as a submodule instead. It currently references the 0.8 branch of
Guichan, after I cherry-picked the CMake support from the master
branch.
When using SDL2 installed through Homebrew, it was necessary to use
SDL2_LINK_LIBRARIES, since the libraries occupy various different
directories.
It no longer seems necessary to include an SDLMain.m, so I deleted it.
Removed obsolete OS X instructions.
Also added instructions for installing dependencies on Fedora.
Removed mentioning of /announce and /who in the README, since they were
removed for tmwAthena.
|
|
Previous code was assuming there would be no gaps in the emote IDs.
Also cleaned up some confusion where the "emote ID" being passed around
in the code was often offset by 1. Now it is only offset in
communication with tmwAthena and when saving the shortcuts.
|
|
It happened to me a few times. Needs checking whether we could instead
make sure we always have a Game instance when receiving this message,
because this way a map mask might get ignored.
Or maybe the map mask just needs to be stored elsewhere.
|
|
* The code defining the message IDs and sizes are now generated by the
tools/protocol.py script in the tmwAthena repository.
* Reduced client version from 20 to 6, because that is currently the
minimum supported version, and any adjustments needed for later likely
still need to be made.
* Removed use of no longer handled messages:
- CMSG_SKILL_USE_BEING
- CMSG_SKILL_USE_POSITION
- CMSG_SKILL_USE_MAP
- SMSG_PARTY_MOVE
- CMSG_WHO_REQUEST
- SMSG_WHO_ANSWER
- SMSG_MVP
- SMSG_BEING_MOVE2
- SMSG_BEING_CHANGE_LOOKS
* Some messages were renamed to match the server side
- CMSG_PLAYER_ATTACK -> CMSG_PLAYER_CHANGE_ACT
- CMSG_PLAYER_RESTART -> CMSG_PLAYER_REBOOT
- SMSG_ADMIN_IP -> SMSG_BEING_IP_RESPONSE
Part of addressing issues #55 and #47, which we now know are about
handling SMSG_PLAYER_HP and SMSG_NPC_COMMAND respectively. The client
will now ignore them (with a warning) instead of crash.
|
|
This stuff was removed from tmwAthena, so it will never be supported.
|
|
* Removed /announce command
* Changed "Kick player" in player pop-up to kick by name
Other messages, like CMSG_ADMIN_HIDE, CMSG_ADMIN_MUTE and
CMSG_ADMIN_LOCAL_ANNOUNCE were already not used by the client. Also,
none of this was implemented for Manaserv.
|
|
Apparently, pkg-config files for physfs were not shipping yet in Ubuntu
18.04. Now it no longer checks that we have at least physfs 3.0, but
older versions should be pretty rare anyway.
This reverts part of ffa8ea13fbf50ad59d9fc4cea71b9aec4aca8bac.
|
|
A faulty version check was disabling music loading entirely.
|
|
* Removing unused includes
* Use member initialization
* Use range-based for loops
* Use nullptr
* Removed no longer used aliases
* Use override
* Don't use else after return
* Use '= delete' to remove implicit members
* Use std::string::empty instead of comparing to ""
|
|
* Introduced a LayoutContext that conveniently allows for relayouting
all rows, or just a single one when it is added. BrowserBox::addRow
no longer relayouts all the rows.
* BrowserLink and LinePart are now merged into a new TextRow struct,
so they can be conveniently dropped when the row limit has been
reached.
* Removed "opaque" option, which was enabled by default but disabled
for all BrowserBox instances.
* Removed "always update" option, and instead start delaying relayouting
automatically when there are a lot of rows (> 100 currently).
* Update window now also has text wrapping enabled.
Closes #50
|
|
Doing some cleanups before working towards optimizing this code.
Removed needless additional wrapping code in BrowserBox::addRow, since
the text will be relayouted anyway.
Simplified layouting code a little. For example, there's no need to keep
track of the number of wrapped lines.
Use more optimal data structures, like an std::deque for the text rows
and a plain std::vector for the line parts. Both have less fragmentation
than an std::list.
|
|
This patch is not exhaustive.
|
|
This just checks that the client compiles on Ubuntu for now.
Also fixed a PhysicsFS linker issue introduced in
ffa8ea13fbf50ad59d9fc4cea71b9aec4aca8bac...
|
|
Wrong format was passed to SDL_RenderReadPixels. Issue introduced in
2c51c98625b225cecfb9628c30d62d4e30f7e3e1.
Closes #52
|
|
|
|
C++17 should be universal enough by now.
This raises the minimum CMake to 3.8, which should also pose no problem since
Ubuntu 18.04 already shipped with CMake 3.10.
C++17's 'if' initialization statement is now used in an efficient
implementation of 'join' for vector<string>, found on StackOverflow.
|
|
Replacing the various OS-specific ways implemented for macOS, Windows
and Linux (based on xmessage).
|
|
It seems the implementation of PhysicsFS RWops shipped requires a
somewhat up-to-date PhysFS (beyond just a missing PHYSFS_DECL macro).
Also simplified CMake files somewhat, enabling new behavior for some
policies that were triggering warnings. And removed the no longer used
FindSDL_gfx.cmake module (SDL_gfx is found using pkg-config since
2c51c98625b225cecfb9628c30d62d4e30f7e3e1).
|
|
Closes #51
|
|
Apparently this was supported before, but support for that was lost in
bd1fdd87eed48ba3ffcc413936d6a6a60a429a97.
Fixes #46
|
|
These checks are unnecessary as deleting a null pointer has no effect.
|
|
|
|
|
|
|
|
Continuation of 2c51c98625b225cecfb9628c30d62d4e30f7e3e1, which had
already removed most of the alpha cache in Image.
|
|
Using the latest implementation from ManaPlus. Previous version was
introduced with 844e9a7a72faca6a212e788a3adc45e17f41dca6 and failed to
"secure" the name due to broken conditions.
|
|
modernize-loop-convert
modernize-deprecated-headers
|
|
Fixes compatibility with SDL2. The old version compiled for whatever
reason, but it would crash when the client tried to play a sound.
|
|
SDL_GetKeyboardState array needs to be indexed by scan codes.
|
|
SDL_GetKeyState used to return an array that is indexed by SDLKey, but
its equivalant, SDL_GetKeyboardState, returns an array that is supposed
to be indexed using SDL_Scancode.
|
|
A custom "Mask" property on a layer or a "foregroundXmask" property on a
map can now be used in combination with the SMSG_MAP_MASK to dynamically
disable certain map layers from the server.
Feature previously seen on ManaPlus and implemented for Mana client for
compatibility.
Also added a ResourceRef class for automating the Resource reference
counting.
Closes #44
|
|
For compatibility with TMW, which has not yet adopted a central
"settings.xml" that includes all the other files.
Also, a "name" attribute with an absolute path is now supported in
addition to the relative "file" attribute, since that is what ManaPlus
supported.
This resolves most of issue #43.
|
