Re: GNU make release candidate 4.1.90 available for download

[Paul Smith]

On Sun, 2016-04-24 at 11:45 +0200, h.becker wrote:
> Build on VMS (V8.3/Alpha) fails with:
> $ @makefile
> ...
> Compiling function...
> 
> # define vfork fork
> ...............^
> %CC-W-MACROREDEF, The redefinition of the macro "vfork" conflicts with 
> a current  definition because one is object-like and the other is 
> function-like.  The redefinition is now in effect.
> at line number 141 in file USR_ODS5:[BECKER_H.make-4^.1^.90]makeint.h;1

Oops that's my bad.  autoconf handles this for me so I don't need that
#define.  I'll fix it.

>           if (vms_report_unix_paths)
> ..............^
> %CC-E-UNDECLARED, In this statement, "vms_report_unix_paths" is not
> declared.

That's a bug in the VMS changes; I'll fix it.

> posixos.c:196:5: error: ‘for’ loop initial declarations are only 
> allowed in C99 or C11 mode
>      for (int i = 0; i < 2; ++i)

I'll fix this and two other places I found that use this syntax.  Man,
it's hard to write in C89 ...

> On VMS (V8.3/Alpha) there is no isblank():
> 
>       if (isblank (i))
> ..........^
> %CC-I-IMPLICITFUNC, In this statement, the identifier "isblank" is
> implicitly declared as a function.

I will fix this.  isblank() is well-defined as space or TAB so there's
really no reason to use this function.  I see it's part of C99 *sigh*.

> However, a simple VMS test fails - I don't see any relation to the
> missing function. Or, as there is so much GNV in this version, are VMS
> lexical functions no longer expected to work?
> 
> $ ty x.mf
> all:
>         write sys$$output "all: ''f$$time()'"

This I can't answer; John Malmberg will have to give details.  However,
I see this in the README.VMS:

>   The default is currently the old behavior when running GNU Make from DCL.
>   In the future this may change.  When running make from GNV Bash the new
>   behavior is the default.

It looks to me like you started your test from GNV Bash?  There should
be a way to get the old behavior even when starting from bash.  A quick
read implies that enabling the GNV$MAKE_OLD_VMS setting may help.  Check
the README.VMS file for more details and verify.

Thanks for testing Hartmut!