compile error for mkString on alpha (PR#332)

[Kurt Hornik]

>>>>> albrecht gebhardt writes:

> Full_Name: Albrecht Gebhardt
> Version: 0.90.0
> OS: osf4.0
> Submission from: (NULL) (143.205.61.73)


> I had to apply the following patch to be able to compile on 
> an alpha with DU 4.0E:
> ###############################################
> --- ./src/main/gram.y.mkString-patch    Tue Nov 23 12:16:29 1999
> +++ ./src/main/gram.y   Tue Nov 23 12:16:59 1999
> @@ -56,7 +56,8 @@
>  SEXP           mkFloat(char *);
>  SEXP           mkInteger(char *);
>  SEXP           mkNA(void);
> -SEXP           mkString(const char *);
> +typedef const char cchar;
> +SEXP           mkString(cchar *);
>  SEXP           mkTrue(void);
 
>  /* Internal lexer / parser state variables */
> @@ -1362,7 +1363,7 @@
>  }
 
 
> -SEXP mkString(const char *s)
> +SEXP mkString(cchar *s)
>  {
>      SEXP t;
 
> ################################################
> Without the patch I got:

> cc -I. -I../include -I../../src/include -I/usr/local/include -DHAVE_CONFIG_H
> -ieee_with_inexact -g -c gram.c -o gram.o
> cc: Warning: ../../../R/src/main/gram.y, line 1626: # not in column 1 is
> ignored, skipping to end of line. (ignoretokens)
>             #if 0
> ------------^
> cc: Warning: ../../../R/src/main/gram.y, line 1628: # not in column 1 is
> ignored, skipping to end of line. (ignoretokens)
>             #endif
> ------------^
> cc: Warning: ../../../R/src/main/gram.y, line 644: In this statement, the
> referenced type of the pointer value "SourceLine" is "unsigned char", which is
> not compatible 
> with "const signed char". (ptrmismatch)
>                     STRING(source)[lines++]  = mkChar(SourceLine);
> ------------------------------------------------------^
> cc: Warning: ../../../R/src/main/gram.y, line 1144: In this statement, the
> referenced type of the pointer value "buf" is "signed char", which is not
> compatible with "un
> signed char". (ptrmismatch)
>                                  buf, 1024, 1) == 0) return R_NilValue;
> ---------------------------------^
> cc: Warning: ../../../R/src/main/gram.y, line 1177: In this statement, the
> referenced type of the pointer value "buf" is "signed char", which is not
> compatible with "un
> signed char". (ptrmismatch)
>                                  buf, 1024, 1) == 0) return R_NilValue;
> ---------------------------------^
> cc: Error: ../../../R/src/main/gram.y, line 1365: In this declaration, parameter
> 1 has a different type than specified in an earlier declaration of this
> function. (mism
> atparam)
> SEXP mkString(const char *s)
> -----^
> cc: Error: ../../../R/src/main/gram.y, line 1365: In this declaration, the type
> of "mkString" is not compatible with the type of a previous declaration of
> "mkString" at
>  line number 59 in file ../../../R/src/main/gram.y. (notcompat)
> SEXP mkString(const char *s)
> -----^

> Then I looked at the preprocessed sources:
> cc -I. -I../include -I../../src/include -I/usr/local/include -DHAVE_CONFIG_H
> -ieee_with_inexact -g -E gram.c > gram.E

> grep -n mkString gram.E 
> 8915:SEXP mkString(const char*);
> 9564:SEXP               mkString(const char *);
> 11995:SEXP mkString( char *s)
> 12179:    protect(yylval = mkString(yytext));

> I don't know why the third occurrence of "mkString(const char*);" 
> (line 11995 of preproccessor output) has the "const" removed, it is present in
> the 
> source! But the trick with typedefing "const char" to "cconst" in the patch
> above works. 


> BTW, there are many warnings about mixing "char" and "unsigned char", 
> not only in gram.c!

> Albrecht

I am not sure that this is the right way of fixing it.  configure has a
macro called AC_C_CONST which checks whether the C compiler fully
supports the `const' keyword.  If not, `const' is defined empty.

Can you look at you Rconfig.h and see if anything is done about const
there?  Can you also look at config.log?

Thanks
-k
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-devel-request@stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._