[PATCH] CParser removed + Hotkey patch

[Jorma Rebane]

1) CParser was a horrible freak of a regex parser. It left a horrible memory footprint and slowed the game down during load times. After replacing it with something much much simpler, the game startup and exit time was noticeably faster (alloc/dealloc overhead probably).

How CTokenizer works:

A string tokenizer doesn't own the string or memory you give it, so make sure the resource exists while you parse.

    std::string input = " toggle = 'Alt+F' = else ";
    std::string key, value;

    CTokenizer tokenizer(input);
    CTokenizer token;

    if(tokenizer.next(token, '=')) // split tokens at '='
    {   // if a token was availablee (no EOS)
        token.trim().toString(key); // trim whitespace, then tostr
        // key == "toggle"
    }
    if(tokenizer.next(token, '='))
    {
        token.trim(" '").toString(value); // trim spaces and '
        // value == "Alt+F"
    }

2) Hotkey patch is a fix that was pretty easy to make - it makes hotkey triggers a lot faster and memory efficient than they used to be. No real functionality changed inside 0 A.D. - only the stuff under the hood got changed.

[Kieran P]

Compiling error on my mac:

In file included from ../../../source/ps/ConfigDB.h:53,
                 from ../../../source/graphics/GameView.cpp:43:
../../../source/ps/Parser.h: In constructor ‘SStrTok::SStrTok(const std::string&)’:
../../../source/ps/Parser.h:62: error: ‘const struct std::basic_string<char, std::char_traits<char>, std::allocator<char> >’ has no member named ‘front’
../../../source/ps/Parser.h:62: error: ‘const struct std::basic_string<char, std::char_traits<char>, std::allocator<char> >’ has no member named ‘back’
../../../source/ps/Parser.h: In constructor ‘SStrTok::SStrTok(const CStr8&)’:
../../../source/ps/Parser.h:63: error: ‘const class CStr8’ has no member named ‘front’
../../../source/ps/Parser.h:63: error: ‘const class CStr8’ has no member named ‘back’
NetServer.cpp
make[1]: *** [obj/graphics_Release/GameView.o] Error 1
make: *** [graphics] Error 2

[Kieran P]

Infact, I'm getting a few errors during compile. Here is a pastie with all of them: ​http://pastie.org/8081539

[Jorma Rebane]

Replying to k776:

Infact, I'm getting a few errors during compile. Here is a pastie with all of them: ​http://pastie.org/8081539

Hi k776, thanks for taking a look at the patch. :) What version of GCC did you compile with? 4.2 or 4.7? std::string::front() and std::string::back() are standard functions, but I can easily replace them with something different.

As for the warnings of 'redundant redefinition of wcsdup', it seems to be some compatibility code in lib/posix/posix.h:

#if OS_MACOSX
# define EMULATE_WCSDUP 1
# define EMULATE_WCSCASECMP 1
#else
# define EMULATE_WCSDUP 0
# define EMULATE_WCSCASECMP 0
#endif

#if EMULATE_WCSDUP
extern wchar_t* wcsdup(const wchar_t* str);
#endif

#if EMULATE_WCSCASECMP
extern int wcscasecmp(const wchar_t* s1, const wchar_t* s2);
#endif

Perhaps you are running a newer compiler that now has wcsdup and wcscasecmp on MAC? In this case we'd have to update the condition:

Edit: Thanks for Markus for pointing out a mistake in the emulation condition. :)

// emulate for < 4.0 or < 4.7
#if OS_MACOSX && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 7))
# define EMULATE_WCSDUP 1
# define EMULATE_WCSCASECMP 1
#else
# define EMULATE_WCSDUP 0
# define EMULATE_WCSCASECMP 0
#endif

[Markus]

Replying to RedFox:

Perhaps you are running a newer compiler that now has wcsdup and wcscasecmp on MAC? In this case we'd have to update the condition:

Thats what #1149 is about.

[Jorma Rebane]

Replying to Markus:

Thats what #1149 is about.

Thanks for pointing that out :) That ticket is 17 months old, though? Can someone test if adding that GNUC check fixes the problem on their MAC? If so, we can patch that ticket, too.

[Jorma Rebane]

Updated the patch file. Removed setlocale(LC_NUMERIC, "C") and atof calls. As a suggestion from leper, just used the simplest atof implementation possible, which should cover our needs.

Also fixed a bug in Hotkey parsing. Now duplicate/secondary hotkey definitions are ignored.

[Jorma Rebane]

Fixed a small parsing bug that popped up after removing C atof. Empty tokens were sometimes not discarded. Using _tofloat to parse floats now; some slight optimizations added.

[Kieran P]

Applied the latest patch and still getting compile errors. Here is build log and errors: ​http://pastie.org/8096337

Compiling with GCC 4.2.1.

[Jorma Rebane]

Replying to k776: Looks like I accidentally left out TerrainManager.cpp. It's already becoming difficult to keep track of which changes apply to which patch... :/

[Jorma Rebane]

Since my patches have started overlapping each-other, its impossible to submit one here. Check the original performance ticket #1995 for the cumulative patch.

[historic_bruno]

Closing these tickets as no further active development is expected. See Philip's megapatch-split git branch for an attempt at splitting megapatch into it's separate parts: ​http://git.wildfiregames.com/gitweb/?p=0ad.git;a=shortlog;h=refs/heads/projects/philip/megapatch-split