[agd-cvs] server BUGS,1.12,1.13 NEWS,1.12,1.13 TODO,1.23,1.24 ChangeLog,1.21,1.22 Ideas,1.9,1.10

[Peep P. <so...@us...>]

Update of /cvsroot/agd/server
In directory sc8-pr-cvs1.sourceforge.net:/tmp/cvs-serv11107

Modified Files:
	BUGS NEWS TODO ChangeLog Ideas 
Log Message:
Bing, changes\!

Index: NEWS
===================================================================
RCS file: /cvsroot/agd/server/NEWS,v
retrieving revision 1.12
retrieving revision 1.13
diff -u -d -r1.12 -r1.13
--- NEWS	7 Jun 2004 15:50:19 -0000	1.12
+++ NEWS	21 Jul 2004 12:13:09 -0000	1.13
@@ -1,9 +1,10 @@
 Changes in 0.0.3:
+	* Arrays!
 	* Renamed interactivep() to interactive()
 	* Locals can now be defined in the middle of code.
 	* dfun platform() changed to a preprocessor macro __ARCH__,
 	  version() to __VERSION__.
-	* New dfuns: throw()
+	* New dfuns: throw(), users()
 	
 Features in 0.0.2-3:
 	* Real string ranges like string[0..4]

Index: BUGS
===================================================================
RCS file: /cvsroot/agd/server/BUGS,v
retrieving revision 1.12
retrieving revision 1.13
diff -u -d -r1.12 -r1.13
--- BUGS	7 Jun 2004 15:31:21 -0000	1.12
+++ BUGS	21 Jul 2004 12:13:09 -0000	1.13
@@ -1,9 +1,7 @@
 Currently known bugs:
-	- src/options and src/compile_options.h don't honor $prefix.
-	  Could use a shell script and sed to do substitutions.
-	  But not that important, I guess.
 	- crashes on Solaris/SPARC with a SIGSEGV somewhere
 	  between lines 190 and 199 in net.c (don't have gdb
 	  on that host). Could this be the wretched inet_ntoa() problem?
 	- on 64-bit, the interpreter fails miserably - I think the code pointer (int *cp) is
 	  incremented by 4 bytes, when it should be incremented by 8.
+	- select() just blocks after a player receives a runtime error

Index: TODO
===================================================================
RCS file: /cvsroot/agd/server/TODO,v
retrieving revision 1.23
retrieving revision 1.24
diff -u -d -r1.23 -r1.24
--- TODO	20 Jun 2004 14:05:16 -0000	1.23
+++ TODO	21 Jul 2004 12:13:09 -0000	1.24
@@ -1,16 +1,40 @@
 Not necessarily in order of importance.
 '!' means first-priority,
-'?' means trivial and not really important.
+'?' means trivial and/or not really important.
 '<' means low priority
 
+ - Runtime token positions (aka line number information).
+   How to do it. Just like in LPMUD. program_t->line_numbers is
+   an array with the same length as there are lines. Each
+   element contains the index into the code array at which
+   the line begins. To find a line number using cp:
+   get offset into code (cp - code), loop through line_numbers
+   while our offset is bigger than the element in line_numbers.
+   If offset is smaller or equal, it means that it must be on
+   the current index.
+ 
+ - profiling memory usage - find out if variables are really freed and if they're
+   being overduplicated or something.
+
+ - everything to do with arrays, notably:
+   - operators:
+     & &=
+   - call_other on an array
+   - foreach()
+
+ - dfuns
+     delete
+	 member_array
+	 explode, implode
+	 save_object, restore_object
+	 crypt
+		
  - allow 'string foo; foo[0]="a";'
  - sprintf and printf
 
  - Interpreter.
    ? F_PUSH_NUM_INT <constant> <number>
-   ? F_POP_N <number> - pops <number> elements off the stack
    < runtime() for dfuns (doesn't display line numbers)
-   < display stack trace (with line numbers and functions) when getting a runtime error
 
  lang.y:
    ! break, continue
@@ -45,16 +69,14 @@
  
  - allow both .c and .lpc for file extensions (/usr/bin/file could recognize LPC,
    and editors can turn on syntax highlighting automatically)
- - Runtime token positions (with the F_NEWLINE operator).
  - keep interactive objects in a separate linked list, and make shout act on that list
 - LPC:
-	! arrays (T_ARRAY)
-	- foreach(), switch()
+	- switch()
 	- all of the operators (/doc/lpc/operators)
 		- ',' - this takes some serious grammar-hacking, because
 		  comma as a function call argument is a separator for expressions.
 	- mixed type
-	- string slices (ranges)
+	- string slices
 		! assigning to slices (F_SLICE_LVALUE etc.)
 		  - Check if assigning too long a string.
 		* m..
@@ -72,32 +94,25 @@
 	     warnings
 	     low-warnings
 	     no-warnings
-	  - #define name value
-	  - #define name(x) value x
-	  - #include, #ifdef, #else, #endif
 	  - #if
 	    - defined()
 	    - have_dfun()
 	    - !, &&, ||
-	  - #elsif
   	  - predefined names:
-	    #define __AGD__ 1
-	    #define __VERSION__ PACKAGE_VERSION
-	     Also possible:
-	       #define __AGD_MAJOR__ 0
-	       #define __AGD_MINOR__ 0
-	       #define __AGD_SUBMINOR__ 2
-	       #define __AGD_PATCHLEVEL__ 3
-	     But these are quite pointless at this time.
-	   #define __ARCH__ PLATFORM
-	   #define __HOST__ the return value of gethostname()
-	   #define __AUTHORS__ ? :)
+	   __HOST__ the return value of gethostname()
+	   __AUTHORS__
 	- character encoding inside strings:
  	  - decimal: \nnn
 	  - octal: \0nnn
 	  - hexadecimal: \0xnn
 	  - binary: \nnnnnnnnb
-	- multiline strings
+	- multiline strings (like this:
+		string s = "@@@FOO
+			blah
+			blah
+			blah
+			FOO";
+		Or however it's done elsewhere.
 
  ? Better function lookup table (prog->nametable)? O(1) would be nice.
  
@@ -106,16 +121,14 @@
  	- log using lib's facilities (master::log()?)
 	- different prefixes have different debuglevels
 	- save to different files
- ? only report undeclared identifiers once (this is generally a Bad Thing)
  - warn about unused variables
  - initialization of global variables
  - dfun overrides, :: operator
+ 
  ? check for lib root and folder of debug log with access() and exit gracefully if it fails.
-
  ? if we crash while running apply(master::crash), then we should rerun with debugging on
+
  ? IPv6 support?
- ? Supply a malloc implementation with statistics (a la MudOS)?
- ? Statistics and profiling
  ? Some work could be done on i18n. For example, allow for multilingual yyerror-messages.
    One way would be a master::error_msg() apply that takes a numeric code of the error
    message and should return the string. Mudlib coders can code in their own language.

Index: Ideas
===================================================================
RCS file: /cvsroot/agd/server/Ideas,v
retrieving revision 1.9
retrieving revision 1.10
diff -u -d -r1.9 -r1.10
--- Ideas	8 Jun 2004 20:18:43 -0000	1.9
+++ Ideas	21 Jul 2004 12:13:09 -0000	1.10
@@ -1,5 +1,9 @@
 This file contains features that will be or will be not in AGD in the future.
 lpc:
+	* The traditional naming conventions of LPC should be rationalized.
+	  this_player() and this_object() are a bit too long to type out
+	  each time. Just player() and this() or self() should suffice.
+	  environment() -> env() etc, too.
 	* superscope global variables. Maybe called world scope?
 	  i.e. global foo; - then can be accessed from any object.
 	  And this_object/this_player are driver-defined world globals.

Index: ChangeLog
===================================================================
RCS file: /cvsroot/agd/server/ChangeLog,v
retrieving revision 1.21
retrieving revision 1.22
diff -u -d -r1.21 -r1.22
--- ChangeLog	20 Jun 2004 14:05:16 -0000	1.21
+++ ChangeLog	21 Jul 2004 12:13:09 -0000	1.22
@@ -1,5 +1,14 @@
 0.0.3:
 ----------------------------------------------------------------------------
+2004-06-23
+	* compile.c,lang.y,interpret.c,vars.c: + operator for arrays
+	* compile.c: fixed adding definitions for prototypes - their code wasn't
+	  added
+	* main.c: startup_time is set before compiling master so that it could
+	  use uptime().
+	* lang.y: fixed popping locals after leaving block
+2004-06-22
+	* interpret.c,lang.y: arrays basically work.
 2004-06-15
 	* interpret.c: TAKE_SLICE is now a function instead of a macro
                    GET_ARGS changed: arg[0] is now the FIRST argument and