--- parser3/INSTALL 2005/12/29 08:27:42 1.48.2.2 +++ parser3/INSTALL 2009/07/29 04:59:15 1.53 @@ -1,13 +1,15 @@ Alexander Petrosian (http://paf.design.ru) -$Id: INSTALL,v 1.48.2.2 2005/12/29 08:27:42 paf Exp $ +$Id: INSTALL,v 1.53 2009/07/29 04:59:15 misha Exp $ 1.What is the process to compile Parser3? Just start ./buildall-with-xml script and you would get $HOME/parser3install/bin/parser3 binary. - In case you do not need XML support, use ./buildall-without-xml NOTE: + In case you do not need XML support, use ./buildall-without-xml + + NOTE: On some platforms there is special version of libgc patched to support some local piece of subtlety. Official libgc does not work there, use special version instead. @@ -15,18 +17,29 @@ $Id: INSTALL,v 1.48.2.2 2005/12/29 08:27 If you need that special version, please change buildall script accordingly. NOTE: + In case you later would experience problems with gcc runtime exception handling + (most notable when reporting sql-related problems), + add this to configure line in buildall script: + --with-sjlj-exceptions + (HPUX is reported to have such problems) + + NOTE: On some systems there are no "make" and you should run 'gmake', change buildall script accordingly then. 2.What is the process to install Parser3? - Just copy files from $HOME/parser3install directory into your cgi-bin directory. + Copy files from $HOME/parser3install directory into your cgi-bin directory. - Files layout: - $prefix/bin/ + Then install Parser3 as a handler of documents, step-by-step instrunctions: + http://www.parser.ru/en/docs/lang/install4apachecgi.htm" in English + http://www.parser.ru/docs/lang/install4apachecgi.htm" in Russian + + Directory layout: + bin/ parser3 -- CGI and command line Parser3 interpreter auto.p.dist -- configuration file sample, copy it to auto.p and adjust to your needs - $prefix/etc/ + etc/ parser3.charsets/ -- charset definition files koi8-r.cfg -- cyrillic charset [KOI8-R encoding] windows-1250.cfg -- central europian charset [windows-1250 encoding] @@ -85,13 +98,6 @@ $Id: INSTALL,v 1.48.2.2 2005/12/29 08:27 option to configure, for example: "--with-sendmail=/usr/sbin/sendmail -i -t -f postmaster" - In case you are experiencing problem with gcc runtime exception handling - (most notable when reporting sql-related problems), - try to configure with - --with-sjlj-exceptions - option. - (HPUX is reported to be have such problems) - By default libstdc++ is linked statically, you can re-enable dynamic libstdc++ linkage by specifying --with-dynamic-stdcpp option to configure. @@ -120,3 +126,48 @@ $Id: INSTALL,v 1.48.2.2 2005/12/29 08:27 WARNING: if you want to use apache configure option --enable-shared=max, do this: ./p3runConfigure --enable-shared=max --disable-shared=mod_parser3.c + +6. + Since Parser 3.4.0 the several optimisations were implemented. + If you experience problems you can try to disable them by commenting corresponding defines and recompiling parser3. + + src/include/pa_opcode.h + #define OPTIMIZE_BYTECODE_GET_CLASS -- $a: + #define OPTIMIZE_BYTECODE_GET_ELEMENT -- $a ^a + #define OPTIMIZE_BYTECODE_GET_OBJECT_ELEMENT -- $a.b ^a.b + #define OPTIMIZE_BYTECODE_GET_OBJECT_VAR_ELEMENT -- $a.$b ^a.$b + #define OPTIMIZE_BYTECODE_GET_SELF_ELEMENT -- $self.a ^self.a + #define OPTIMIZE_BYTECODE_CONSTRUCT -- $a(expr), $a[value] + $.a(expr), $.a[value] + $self.a(expr), $self.a[value] + #define OPTIMIZE_BYTECODE_CONSTRUCT_OBJECT -- ^class::constructor + #define OPTIMIZE_BYTECODE_CUT_REM_OPERATOR -- cut rem operator with any number of params during compilation + #define OPTIMIZE_BYTECODE_STRING_POOL -- simplifying string's bytecode into expression + + src/include/pa_string.h + #define HASH_CODE_CACHING -- calculated hash codes are cached and used for sequential hash lookups + + src/include/pa_memory.h + #define USE_DESTRUCTORS -- destructors are used to decrease memory consumption during code + processing and make ^memory:compact[] calls non-essential. + + src/types/pa_method.h + #define OPTIMIZE_CALL -- allows faster operators execution by eliminating method frame + with local variables creation and extra write context switches. + #define OPTIMIZE_RESULT -- parser methods are marked when the $result variable was used. + This allows write operations and context switching optimization. + + src/types/pa_wwrapper.h + #define OPTIMIZE_SINGLE_STRING_WRITE -- reuse original VString in single string assignments. + + src/lib/cord/include/cord.h + #define CORD_CAT_OPTIMIZATION -- CORD library never modifies source concatenations. But in parser + write operations it is safe to modify them and save some memory. + #define CORD_CHARS_CACHE -- language cords with same language and length are cached and reused. + + src/include/pa_string.h + #define STRING_LENGTH_CACHING -- cache String::Body.length() for char* strings + + src/include/pa_hash.h + #define HASH_ORDER -- hash keys are iterated in the order of insertion +