Commit Graph

6187 Commits

Author SHA1 Message Date
pierre
6800fa46bc * avoid cyclic overloaded list for operators 2000-04-03 14:50:05 +00:00
pierre
7f4168fb0a * fix for bug #909 2000-04-03 12:23:02 +00:00
michael
620e7fc689 * TRim(right|Left) more Delphi compatible 2000-04-03 06:40:37 +00:00
florian
a2e8fd9f07 *** empty log message *** 2000-04-02 20:02:00 +00:00
florian
ff5b72e3bd * fixed another problem with readln(<floating point register variable>);
* the register allocator takes now care of necessary pushes/pops for
    readln/writeln
2000-04-02 18:30:11 +00:00
florian
e6f4922ad6 *** empty log message *** 2000-04-02 17:49:09 +00:00
florian
f71ddb9e2f * readln(r); works now, if r is a fpu register variable 2000-04-02 17:47:47 +00:00
florian
3f9126f999 * fixed bug #903: the compiler gives now a nice message if it can't create
the .o file, (same for future .ar)
2000-04-02 15:22:17 +00:00
florian
b8faf9a8a3 * some more procedures can be now hooked by the OS specific implementation 2000-04-02 12:13:36 +00:00
florian
2c210cd7ee * a lot of optimizing done 2000-04-02 12:12:22 +00:00
florian
7c75febd64 * enumerations with size 1 or 2 weren't handled corretly if they were register
variables: in fact they got 32 bit register assigned; fixed
2000-04-02 12:11:38 +00:00
florian
099f13ec92 *** empty log message *** 2000-04-02 12:10:11 +00:00
florian
38567989b0 *** empty log message *** 2000-04-02 11:35:48 +00:00
florian
339d986a2a * bug #701 fixed: ansistrings in interface and implementation part of the units
are now finalized correctly even if there are no explicit initialization/
    finalization statements
2000-04-02 10:18:18 +00:00
florian
b16a589d24 * proper test for web bug #701 2000-04-02 09:39:54 +00:00
florian
548f14e16d * halt in the finalization statement of a unit lead to an endless loop; fixed 2000-04-02 09:39:25 +00:00
florian
03052cb67e *** empty log message *** 2000-04-02 09:37:55 +00:00
florian
fcf5315f23 + constant procedure variables can have a @ in front:
const p : procedure = @p;
    til now only
       const p : procedure = p;
    was allowed
2000-04-02 09:12:51 +00:00
florian
3dfffb4d75 *** empty log message *** 2000-04-02 09:06:55 +00:00
michael
e02f1fc294 + Initial implementation 2000-04-01 20:56:30 +00:00
michael
f097333d67 + Added cmem unit 2000-04-01 20:49:37 +00:00
michael
d66417a0df + Initial implementation 2000-04-01 20:48:43 +00:00
florian
deb19f0eb6 + implemented support for 64 bit int types
+ options are written now to output file
  * improved compact mode
  * fixed writing of variables
2000-04-01 20:19:37 +00:00
michael
a9a42cb381 + Implemented ibase 6.0 2000-04-01 19:18:52 +00:00
michael
8d37e95c9e + Implemented ibase 6.0 2000-04-01 18:42:15 +00:00
peter
836eafa712 * use arraydef.elesize instead of elementtype.def.size 2000-04-01 14:18:44 +00:00
peter
8550915222 * don't write suffix for fldcw 2000-04-01 14:18:03 +00:00
peter
0fccba3864 * arraydef.elesize returns 4 when strings are found in an openarray,
arrayconstructor. Since only the pointers to the strings are stored
2000-04-01 14:17:08 +00:00
peter
a69a6856be * addition for another procvar style decl (not working correct yet) 2000-04-01 14:16:31 +00:00
peter
8959ab2eda * mode H+ added 2000-04-01 11:54:04 +00:00
peter
96a0b5cd32 * dependencies added 2000-04-01 11:50:39 +00:00
peter
7c516d9404 * fixed addref/decref for records, the data/type was read in the
wrong order
2000-04-01 11:45:34 +00:00
peter
7c04f2ff09 * fixed rtti info for record 2000-04-01 11:44:56 +00:00
Tomas Hajny
b65e936773 * logfile appended if exists 2000-04-01 10:46:29 +00:00
Tomas Hajny
4909e55df6 * DosGetInfoBloBlocks updated 2000-04-01 10:46:07 +00:00
Tomas Hajny
9e0a0b54c7 OS/2 implementation started 2000-04-01 10:45:52 +00:00
Tomas Hajny
f295a8029b * .ao2 bug fixed 2000-04-01 10:45:14 +00:00
pierre
997b0d9f5b * FPU needs reset for all SIGFPE even from integer division by zero 2000-03-31 23:26:32 +00:00
pierre
74907e307a * multiple exception handling works
(for linux only if syslinux is compiled with -dnewsignal)
2000-03-31 23:21:19 +00:00
pierre
f6da28673a * changed handling of interrupt 0x75 :
the status word is saved into ___djgpp_fpu_state
    and inserted in __sigmaks field of djgpp exception record
    BUT fnclex is called after to avoid a second interrupt
    generation on fn??? calls
2000-03-31 23:19:12 +00:00
pierre
beecf1ec51 * remove fninit in longjump 2000-03-31 23:12:09 +00:00
pierre
77c220c69a * TSigAction Context param is the full record not a pointer 2000-03-31 23:11:23 +00:00
pierre
92e9d3bf83 * fix the handling of value parameters in cdecl function 2000-03-31 22:56:45 +00:00
Jonas Maebe
cc86370fa3 * signal handling using sigaction when compiled with -dnewsignal
(allows multiple signals to be received in one run)
2000-03-31 13:24:28 +00:00
pierre
8ad66dbf27 * use $(AS) instead of as for cross compiling 2000-03-31 12:46:16 +00:00
Tomas Hajny
531cb20c02 * DIVE updated 2000-03-30 18:41:50 +00:00
pierre
ad052390fb No stack check inside CatchUnhandledException 2000-03-30 13:54:15 +00:00
pierre
d74dc01fef * fix FPU and multiple exception problems 2000-03-30 13:40:57 +00:00
pierre
973e5390c8 check for multi times FPU exception generation 2000-03-30 12:58:02 +00:00
pierre
f06a17c7c7 * Require error 201 with range check on 2000-03-29 08:05:13 +00:00