$ONEXIT

[jcfuller]

Kevin,
  A small issue with $ONEXIT.
I use it in all my code to run my compile batch files.
It works and every thing compiles fine but the last line seems out of place?
James

Code Select Expand


$ONEXIT "NUWENGPP.BAT $FILE$ con"

bc "C:\BcxRadAsm3\RadAsm\Bcx\Projects\test01\test01"

BCX BASIC to C/C++ Translator (c) 1999 by Kevin Diggins
Version 7.4.0 (2020/01/19)
[Lines In: 34] [Lines Out: 446] [Statements: 14] [Time: 0.04 sec's]
BCX translated C:\BcxRadAsm3\RadAsm\Bcx\Projects\test01\test01.bas to C:\BcxRadAsm3\RadAsm\Bcx\Projects\test01\test01.cpp For a C++ Compiler

Shelling Out To:NUWENGPP.BAT C:\BcxRadAsm3\RadAsm\Bcx\Projects\test01\test01 con
Compiling "test01.cpp" to a  64bit "Windows Console App" Using NUWEN Distro
test01.cpp: In function 'int scan(char*, char*, ...)':test01.cpp:359:21: warning: ISO C++ forbids converting a string constant to 'char*' [-Wwrite-strings]  359 |   d = Split(A,input,",");      |                     ^~~test01.cpp: In function 'int main(int, char**)':test01.cpp:465:30: warning: ISO C++ forbids converting a string constant to 'char*' [-Wwrite-strings]  465 | ScanError = scan(InputBuffer,"%d",&K);      |                              ^~~~Finished!
Can't open file C:\BcxRadAsm3\RadAsm\Bcx\Projects\test01\test01



[MrBcx]

James - if you email me your batch file and test01, I'll take a look. 

[MrBcx]

That was tricky but I believe I found the cause and, more importantly, the fix.

At the bottom of SUB PostProcess,  add the three (3) new lines of code (in bold) shown below:
  DIM szSave$
  szSave$ = FileOut$
.
.
.
  FileOut$  = szSave$
END SUB ' PostProcess

Code Select Expand


  '**************************
  DIM szSave$
  szSave$ = FileOut$
   
  FOR INTEGER A = 1 TO XitCount
    FileOut$ = EXTRACT$(FileOut$,".")
    Cmd$ = REMOVE$(Xit$[A],DQ$)
    IREPLACE "$FILE$" WITH OutfileClone$ IN Cmd$
    IREPLACE "$PELLES$\\" WITH PELLESPATH$ IN Cmd$
    IREPLACE "$PELLES$"   WITH PELLESPATH$ IN Cmd$
    IREPLACE "$LCC$\\" WITH LCCPATH$ IN Cmd$
    IREPLACE "$LCC$"   WITH LCCPATH$ IN Cmd$
    IREPLACE "$BCX$\\" WITH BCXPATH$ IN Cmd$
    IREPLACE "$BCX$"   WITH BCXPATH$ IN Cmd$
    REPLACE "\\\\" WITH "\\" IN Cmd$
    REPLACE DDQ$ WITH DQ$ IN Cmd$
    PRINT "Shelling Out To:", Cmd$
    SHELL Cmd$   
  NEXT
  FileOut$  = szSave$
END SUB ' PostProcess


[jcfuller]

Kevin,
  After I set up a bcx compile folder I compiled with Pelles C 32 and it ran fine.
BUT when I did a compile setup similar to my bc9 using TDM g++ I relived the HORRORS of:
warning: ISO C++ forbids converting a string constant to 'char*' [-Wwrite-strings]
I experienced when I first started bc9.
I bit the bullet and removed every single warning in bc9. Are you up to the task

James

[MrBcx]

Being up to the task and being committed to the task are two different things.

I'd have to say that there is about a 0.1% likelihood that I'll spend any time on such an endeavor.

My attitude about warnings is that they mean nothing until they mean something.

[jcfuller]

Being up to the task and being committed to the task are two different things.

I'd have to say that there is about a 0.1% likelihood that I'll spend any time on such an endeavor.

My attitude about warnings is that they mean nothing until they mean something.

Kevin,
I thought I would give a go at fixing the warning:
ISO C++ forbids converting a string constant to 'char*' [-Wwrite-strings]
when it dawned on me I really don't care when compiling Bcx. I can use -Wno-write-strings.
But I don't want MY translated code to spout them from the runtime code produced by Bcx.
I thought I might have better luck dumping the rtl and focus on converting those functions and it would be easier to update those in the Bcx source itself.

Alas I could not get bc -l to work??

James

[Robert]

Alas I could not get bc -l to work??

James

Hi James:

bc -l works for me with BCX 7.4.1.

Note well, the rtlib and other folders are created in the root directory of the drive from which "bc -l" is invoked.

I invoked bc -l from C:dev\ and the rtlib and other folders were created in C:\

[jcfuller]

Thanks Robert. Found them
James

[MrBcx]

I don't know whether this will interest you but I'll offer it.

I wrote a tool to help me analyze BCX by breaking out the key components.


Rather than bore you with the details of my tool, I am attaching the output as a zip file for your examination.

If you translate Bc_(1)_NEW_BUILD.Bas and compile, you will have recreated the BCX translator.
That's not the point ...
The point is that successfully compiling all the parts gives confidence in BINK's disassembly.


C:\Temp\BCX_By_The_Numbers>bink bc.bas
Opening Bc.Bas
Reading Bc.Bas
Closing Bc.Bas

Creating Bc_(1)_NEW_BUILD.BAS

Found  152 SUBS
Found  140 FUNCTIONS
Found  13 ENUMS
Found  11 SETS
Found  15 TYPES
Found  92 CONSTANTS
Found  595 GLOBALS

[jcfuller]

Kevin,
   With such a  tools search capability why can't you tweak it to change all the char* to const char* in the runtime functions output in bc.bas itself?

James

[MrBcx]

James ... give me a before and after example of what you're talking about.

[jcfuller]

Kevin,
  I believe this all you need to eliminate all the warnings when you compile as c++ source.

James



Before:

Code Select Expand


    IF Use_Str_Cmp THEN
      FPRINT FP_WRITE, "int     str_cmp(char*, char*);"
    END IF

............

  IF Use_Str_Cmp THEN
    IF Use_Library THEN FPRINT FP_WRITE, "// BCXRTLIB: str_cmp"
    FPRINT FP_WRITE, "int str_cmp (char *a, char *b)"
    FPRINT FP_WRITE, "{"
    FPRINT FP_WRITE, "  int counter=0;"
    FPRINT FP_WRITE, "  for(;;)"
    FPRINT FP_WRITE, "   {"
    FPRINT FP_WRITE, "    if((a[counter]^b[counter]))"
    FPRINT FP_WRITE, "     {"
    FPRINT FP_WRITE, "      if((UINT) a[counter]>= (UINT) b[counter])"
    FPRINT FP_WRITE, "      return  1;"
    FPRINT FP_WRITE, "      return -1;"
    FPRINT FP_WRITE, "     }"
    FPRINT FP_WRITE, "    if(!a[counter]) return 0;"
    FPRINT FP_WRITE, "    counter++;"
    FPRINT FP_WRITE, "   }"
    FPRINT FP_WRITE, sENDBCXRTLIB$
  END IF



After

Code Select Expand


    IF Use_Str_Cmp THEN
      FPRINT FP_WRITE, "int     str_cmp(const char*, const char*);"
    END IF
................................

  IF Use_Str_Cmp THEN
    IF Use_Library THEN FPRINT FP_WRITE, "// BCXRTLIB: str_cmp"
    FPRINT FP_WRITE, "int str_cmp (const char *a, const char *b)"
    FPRINT FP_WRITE, "{"
    FPRINT FP_WRITE, "  int counter=0;"
    FPRINT FP_WRITE, "  for(;;)"
    FPRINT FP_WRITE, "   {"
    FPRINT FP_WRITE, "    if((a[counter]^b[counter]))"
    FPRINT FP_WRITE, "     {"
    FPRINT FP_WRITE, "      if((UINT) a[counter]>= (UINT) b[counter])"
    FPRINT FP_WRITE, "      return  1;"
    FPRINT FP_WRITE, "      return -1;"
    FPRINT FP_WRITE, "     }"
    FPRINT FP_WRITE, "    if(!a[counter]) return 0;"
    FPRINT FP_WRITE, "    counter++;"
    FPRINT FP_WRITE, "   }"
    FPRINT FP_WRITE, sENDBCXRTLIB$
  END IF


[MrBcx]

So ... only the arguments need to be updated ?

I concluded after reading this ( https://stackoverflow.com/questions/9834067/difference-between-char-and-const-char   )
that functions returning "char *" would also need to be updated.  And GLOBAL variables defined as "char *" would need to be updated.

Why be concerned that function arguments might get hijacked by n'er-do-wells but not concerned that functions or variables are equally vulnerable?


This is as clear as mud to me.

[jcfuller]

Kevin,
  I misspoke on the warnings.  My concern is with the rt functions not the bc translator itself at this point. I bit the bullet and eliminated all the warnings in bc9 but it was not easy. It came from compiling over and over until I got no more warnings.

If you want to completely update bc this type will also need to be updated.
As will other items I do not remember.

James

Code Select Expand


SET atVARIANTVALUENAMES[] AS tagVVN
  {"bVal"      , vt_BYTE           , 0},
  {"bool"      , vt__VARIANT_BOOL  , 0},
  {"boolVal"   , vt_VARIANT_BOOL   , 0},
  {"bstrVal"   , vt_BSTR           , 0},
  {"byref"     , vt_PVOID          , 0},
  {"cVal"      , vt_CHAR           , 0},
  {"cyVal"     , vt_CY             , 0},
  {"date"      , vt_DATE           , 0},
  {"dblVal"    , vt_DOUBLE         , 0},
  {"fltVal"    , vt_FLOAT          , 0},
  {"iVal"      , vt_SHORT          , 0},
  {"intVal"    , vt_INT            , 0},
  {"lVal"      , vt_LONG           , 0},
  {"llVal"     , vt_LLONG          , 0},
  {"parray"    , vt_SAFEARRAY      , 1},
  {"pbVal"     , vt_BYTE           , 1},
  {"pbool"     , vt__VARIANT_BOOL  , 1},
  {"pboolVal"  , vt_VARIANT_BOOL   , 1},
  {"pbstrVal"  , vt_BSTR           , 1},
  {"pcVal"     , vt_CHAR           , 1},
  {"pcyVal"    , vt_CY             , 1},
  {"pdate"     , vt_DATE           , 1},
  {"pdblVal"   , vt_DOUBLE         , 1},
  {"pdecVal"   , vt_DECIMAL        , 1},
  {"pdispVal"  , vt_IDispatch      , 1},
  {"pfltVal"   , vt_FLOAT          , 1},
  {"piVal"     , vt_SHORT          , 1},
  {"pintVal"   , vt_INT            , 1},
  {"plVal"     , vt_LONG           , 1},
  {"pllVal"    , vt_LLONG          , 1},
  {"pparray"   , vt_SAFEARRAY      , 2},
  {"ppdispVal" , vt_IDispatch      , 2},
  {"ppunkVal"  , vt_IUnknown       , 2},
  {"pscode"    , vt_SCODE          , 1},
  {"puiVal"    , vt_USHORT         , 1},
  {"pulVal"    , vt_ULONG          , 1},
  {"pullVal"   , vt_ULONGLONG      , 1},
  {"punkVal"   , vt_IUnknown       , 1},
  {"pvarVal"   , vt_VARIANT        , 1},
  {"scode"     , vt_SCODE          , 0},
  {"uiVal"     , vt_USHORT         , 0},
  {"uintVal"   , vt_UINT           , 0},
  {"ulVal"     , vt_ULONG          , 0},
  {"ullVal"    , vt_ULONGLONG      , 0},
  {NULL        , 0                 , 0}
END SET



to

Code Select Expand


SET atVARIANTVALUENAMES[] AS tagVVN
  {(char*)"bVal"     , vt_BYTE         , 0},
  {(char*)"bool"     , vt__VARIANT_BOOL, 0},
  {(char*)"boolVal"  , vt_VARIANT_BOOL , 0},
  {(char*)"bstrVal"  , vt_BSTR         , 0},
  {(char*)"byref"    , vt_PVOID        , 0},
  {(char*)"cVal"     , vt_CHAR         , 0},
  {(char*)"cyVal"    , vt_CY           , 0},
  {(char*)"date"     , vt_DATE         , 0},
  {(char*)"dblVal"   , vt_DOUBLE       , 0},
  {(char*)"fltVal"   , vt_FLOAT        , 0},
  {(char*)"iVal"     , vt_SHORT        , 0},
  {(char*)"intVal"   , vt_INT          , 0},
  {(char*)"lVal"     , vt_LONG         , 0},
  {(char*)"llVal"    , vt_LLONG        , 0},
  {(char*)"parray"   , vt_SAFEARRAY    , 1},
  {(char*)"pbVal"    , vt_BYTE         , 1},
  {(char*)"pbool"    , vt__VARIANT_BOOL, 1},
  {(char*)"pboolVal" , vt_VARIANT_BOOL , 1},
  {(char*)"pbstrVal" , vt_BSTR         , 1},
  {(char*)"pcVal"    , vt_CHAR         , 1},
  {(char*)"pcyVal"   , vt_CY           , 1},
  {(char*)"pdate"    , vt_DATE         , 1},
  {(char*)"pdblVal"  , vt_DOUBLE       , 1},
  {(char*)"pdecVal"  , vt_DECIMAL      , 1},
  {(char*)"pdispVal" , vt_IDispatch    , 1},
  {(char*)"pfltVal"  , vt_FLOAT        , 1},
  {(char*)"piVal"    , vt_SHORT        , 1},
  {(char*)"pintVal"  , vt_INT          , 1},
  {(char*)"plVal"    , vt_LONG         , 1},
  {(char*)"pllVal"   , vt_LLONG        , 1},
  {(char*)"pparray"  , vt_SAFEARRAY    , 2},
  {(char*)"ppdispVal", vt_IDispatch    , 2},
  {(char*)"ppunkVal" , vt_IUnknown     , 2},
  {(char*)"pscode"   , vt_SCODE        , 1},
  {(char*)"puiVal"   , vt_USHORT       , 1},
  {(char*)"pulVal"   , vt_ULONG        , 1},
  {(char*)"pullVal"  , vt_ULONGLONG    , 1},
  {(char*)"punkVal"  , vt_IUnknown     , 1},
  {(char*)"pvarVal"  , vt_VARIANT      , 1},
  {(char*)"scode"    , vt_SCODE        , 0},
  {(char*)"uiVal"    , vt_USHORT       , 0},
  {(char*)"uintVal"  , vt_UINT         , 0},
  {(char*)"ulVal"    , vt_ULONG        , 0},
  {(char*)"ullVal"   , vt_ULONGLONG    , 0},
  {NULL       ,0                , 0}
END SET



[jcfuller]

Kevin,
  Followup: When I compile bcx with the same TDM compiler I us with bc9 using -Wno-write-strings there are no warnings.
Attached is the without.

James