parsed as unsigned constants (otherwise, $80000000 got sign extended
and became $ffffffff80000000), all constants in the longint range
become longints, all constants >$7fffffff and <=cardinal($ffffffff)
are cardinals and the rest are int64's.
* added lots of longint typecast to prevent range check errors in the
compiler and rtl
* type casts of symbolic ordinal constants are now preserved
* fixed bug where the original resulttype wasn't restored correctly
after doing a 64bit rangecheck
associated errors (fillchar(modeinfo,sizeof(tmodeinfo),#0) instead
of sizeof(TVesamodeinfo) etc)
* changed several sizeof(type) to sizeof(varname) to avoid similar
errors in the future
* isgraphmode stuff cleanup
* vesainfo.modelist now gets disposed in cleanmode instead of in
closegraph (required moving of some declarations from vesa.inc to
new vesah.inc)
* queryadapter gets no longer called from initgraph (is called from
initialization of graph unit)
* bugfix for notput in 32k and 64k vesa modes
* a div replaced by / in fillpoly
* closegraph: exits now immidiately if isgraphmode = false (caused
RTE 204 with VESA enabled if you set exitproc to call closegraph
and also called closegraph explicitely before exit, like bgidemo)
* GetScanLine: only get line inside viewport, data outside of it
is not used anyway
* InternalEllipseDefault: fix for when xradius or yradius = 0 and
increase xradius and yradius always by one (TP does this too)
* fixed conlict in vesa.inc from last update
* some conditionals to avoid range check and overflow errors in
places where it doesn't matter
* added first FPC support, only VGA works to some extend for now
* use -dasmgraph to use assembler routines, otherwise Pascal
equivalents are used
* use -dsupportVESA to support VESA (crashes under FPC for now)
* only dispose vesainfo at closegrph if a vesa card was detected
* changed int32 to longint (int32 is not declared under FPC)
* changed the declaration of almost every procedure in graph.inc to
"far;" becquse otherwise you can't assign them to procvars under TP
real mode (but unexplainable "data segnment too large" errors prevent
it from working under real mode anyway)
