[gforth] / gforth / except.fs  

gforth: gforth/except.fs

Diff for /gforth/except.fs between version 1.8 and 1.9

version 1.8, Sun Nov 23 22:02:46 2003 UTC version 1.9, Sun Nov 23 22:52:29 2003 UTC
Line 33 
Line 33 
 ' noop IS store-backtrace  ' noop IS store-backtrace
 \ [THEN]  \ [THEN]
   
   \ Ok, here's the story about how we get to the native code for the
   \ recovery code in case of a THROW, and why there is all this funny
   \ stuff being compiled by TRY and RECOVER:
   
   \ Upon a THROW, we cannot just return through the ordinary return
   \ address, but have to use a different one, for code after the
   \ RECOVER.  How do we do that, in a way portable between the various
   \ threaded and native code engines?  In particular, how does the
   \ native code engine learn about the address of the native recovery
   \ code?
   
   \ On the Forth level, we can compile only references to threaded code.
   \ The only thing that translates a threaded code address to a native
   \ code address is docol, which is only called with EXECUTE and
   \ friends.  So we start the recovery code with a docol, and invoke it
   \ with PERFORM; the recovery code then rdrops the superfluously
   \ generated return address and continues with the proper recovery
   \ code.
   
   \ At compile time, since we cannot compile a forward reference (to the
   \ recovery code) as a literal (backpatching does not work for
   \ native-code literals), we produce a data cell (wrapped in AHEAD
   \ ... THEN) that we can backpatch, and compile the address of that as
   \ literal.
   
   \ Overall, this leads to the following resulting code:
   
   \   ahead
   \ +><recovery address>-+
   \ | then               |
   \ +-lit                |
   \   (try)              |
   \   ...                |
   \   (recover)          |
   \   ahead              |
   \   docol: <-----------+
   \   rdrop
   \   ...
   \   then
   \   ...
   
   \ !! explain handler on-stack structure
   
 : (try) ( ahandler -- )  : (try) ( ahandler -- )
     r>      r>
     swap >r \ recovery address      swap >r \ recovery address
Line 48 
Line 91 
 : try ( compilation  -- orig ; run-time  -- ) \ gforth  : try ( compilation  -- orig ; run-time  -- ) \ gforth
     \ !! does not work correctly for gforth-native      \ !! does not work correctly for gforth-native
     POSTPONE ahead here >r >mark 1 cs-roll POSTPONE then      POSTPONE ahead here >r >mark 1 cs-roll POSTPONE then
     r> POSTPONE literal POSTPONE @ POSTPONE (try) ; immediate compile-only      r> POSTPONE literal POSTPONE (try) ; immediate compile-only
   
 : (recover) ( -- )  : (recover) ( -- )
     \ normal end of try block: restore handler, forget rest      \ normal end of try block: restore handler, forget rest
Line 64 
Line 107 
 : recover ( compilation  orig1 -- orig2 ; run-time  -- ) \ gforth  : recover ( compilation  orig1 -- orig2 ; run-time  -- ) \ gforth
     \ !! check using a special tag      \ !! check using a special tag
     POSTPONE (recover)      POSTPONE (recover)
     POSTPONE else ; immediate compile-only      POSTPONE else
       docol: here 0 , 0 , code-address! \ start a colon def
       postpone rdrop                    \ drop the return address
   ; immediate compile-only
   
 : endtry ( compilation  orig -- ; run-time  -- ) \ gforth  : endtry ( compilation  orig -- ; run-time  -- ) \ gforth
     POSTPONE then ; immediate compile-only      POSTPONE then ; immediate compile-only
Line 91 
Line 137 
         r> lp!          r> lp!
         r> fp!          r> fp!
         r> swap >r sp! drop r>          r> swap >r sp! drop r>
         rdrop 'throw          rdrop 'throw r> perform
     THEN ;      THEN ;
 is throw  is throw
   

Generate output suitable for use with a patch program
Legend:
Removed from v.1.8  
changed lines
  Added in v.1.9
CVS Admin

Powered by ViewCVS 1.0-dev 		(Powered by ViewCVS)

ViewCVS and CVS Help