* Remove prebuilt nasm-generated object files from the tree, and use AMBuild to compile them.
* Add commands in VS project to build the object files
* Rename elf to elf32 for consistency
Modules are loaded before plugins, and can hook OnPluginsLoaded. Therefore, they should be unloaded AFTER plugins, and be able to hook OnPluginsUnloaded. This was not the case for reloadable modules.
This affects nvault module, whose OnPluginsUnloaded function wasn't being called.
* Add JSON module
* Merge upstream changes
Fix memory leaks
* Add json include to PackageScript
* Merge upstream changes
Fix memory leaks and increase max nesting
* Fix documentation
* Use AutoPtr in managing JSON handles
* Merge upstream changes
Order of items in an array is preserved after removing an item.
* Merge upstream
* Fix crash
* Add VS projects files and fix mixed tab/spaces
* Remove erroring on "json_free"
* Add comments to "json.inc" file
* Remove overloaded operators
* Use of "override" keyword where needed
* Fix parameter's name
* Compile as static library, update AMBuildScript and link to core
* Update VS project files to include the library
* Add UTF-8 Rewind library (v1.5.1) to third_party directory
* Update ACKNOWLEDGEMENTS.txt
* Move AMXX buffer in its own function
* Move constants from string.inc to string_const.inc and update project files
* Move stocks from string.inc to string_stocks.inc and update project files
* Improve UTF-8 support in containi() and update documentation
* Improve UTF-8 support in strcmp() and update documentation
* Improve UTF-8 support in strfind() and update documentation
Worth to be noted that this native with ignorecase set was not working properly. So broken that no one reported the issue.
This adds also a safety check for "pos" parameter to not go < 0.
* Improve UTF-8 support in strncmp() and update documentation
* Improve UTF-8 support in equali() and update documentation
* Add an option to some UTF-8 Rewind functions for avoiding invalid data to be replaced
By default it replaces any invalid byte or sequence of bytes by 0xFFFD (3 bytes). It can be problematic when the input buffer is not changed (from a plugin) and that some natives need to calculate a position from the converted string. With such replacement, the position is displaced due the final string length being larger.
This compiles the library as C++, because I added some silly param with a default default value which is not supported by C.
* Improve UTF-8 support in replace_string/ex() and update documentation
* Add is_string_category() and update documentation
* Update a little testsuite plugin (and fix linux compilation)
* Add mb_strotolower/upper() and update documentation
* Add mb_ucfirst() and update documentation
* Add mb_strtotile() and update documentation
* Improve UTF-8 support in get_players() and find_player() with name/case insenstive flags set
* Fix KliPPy's complain
* elog_message
Like "log_message". Logs a message to the current server log file, however the messages sent are also sent to all plugins and can be hooked using "register_logevent".
* Update amxmodx.cpp
* Update Editor.sci
* Update UnitfrmMain.dfm
* Add ReHLDS API files and its dependencies
Note: This has been stolen from ReAPI AMXX module and modified/adjusted to match AMXX existing includes and to provide as less dependencies as possible as well
* Add the necessary files to get ReHLDS interface
* Split SV_DropClient into pre/post code
* Init ReHLDS API and add SV_DropClient hook
* Add Cvar_DirectSet hook and adjust code with helpers
Note: we don't need to split code here. This is pretty much the naive and straight way, but fairly enough for our case. If it happens we got a lot more hooks, we may consider to use some class to manage better the things.
* Move platform and interface stuff in their own files in public directory
* Make sure to init cvar stuff after ReHLDS
* Add ReGameDLL API files and its dependencies in cstrike module
* Init ReHLDS in cstrike module and adjust code
Note: About cs_uset_set_model(). ReHLDS API doesn't offer a way to know directly the precached models, so instead of looping through all the ressources, the models list is saved one time at map change into a hashmap.
* Init ReGameDLL and adjust code
* Fix linux compilation
* Init ReGameDLL in fakemeta module and adjust code
* Rename /reapi directory to /resdk to avoid confusion
* Retrieve gamerules pointer through InstallGameRules in fakemeta module
* Retrieve gamerules pointer through InstallGameRules in cstrike module
Note: actually gamerules is not used if regamedll is enabled, but it could be used in future natives.
* Fix a typo when ReGameDLL is not enabled
* Fix missing interface check for ReHLDS.
I'm pretty sure I was checking at the very first since I worked first on vanilla version of engine, looks like change has been lost.
* Rename client_disconnected to client_disconnecting
* Add client_disconnected as post forward
* Fix client not properly disconnected internally
Introduced in #264.
Edict is reset once SV_DropClient is called, so that second check would be always false.
* Reflect changes on the concerned plugins
* Revert renaming, let's add only client_remove as post forward
Upcasting of -ve signed characters (char) to signed integers (cell) results in -ve signed integers which don't represent valid printable characters.
eg. UTF-8 Character '中' (0xE4 0xB8 0xAD) when casted results in 0xFFFFFFE4 0xFFFFFFB8 0xFFFFFFAD which are not valid printable characters.
* Revert "Extend "read_argv" native"
This reverts commit aaa2934595.
This broke binary compatibility with either older, already compiled
plugins, or newly compiled plugins if you forgot to update .inc headers.
This happened because read_argv used to receive maxlen by value, and
after
this commit it receives it by reference. This causes read_argv either
to fail, or worse, to buffer overflow, resulting in a security vulnerability.
Newly introduced functionality for read_argv should be added with a new,
separate native.
* Add "read_argv_int" & "read_argv_float" natives
* Fix floatround overflow
floatround would overflow for floats greater than 2^30 because internally
it would double the number (therefore anything greater than 2^30 results
in something greater than 2^31 which would cause overflow of course)
floatround behaviour is left exactly identical otherwise (although I find
it very weird and wrong to deliberately avoid banker's rounding, it would
be a bad idea to change this behaviour due to compatibility)
Remember to reassemble amxexecn and amxjitsn
* Update the compiled object files
