Re: make 3.81rc1 / MSYS

[Eli Zaretskii]

> Date: Sat, 25 Mar 2006 10:24:33 -0500
> Cc: David Ergo <address@hidden>, address@hidden,
>    address@hidden, address@hidden
> From: "Paul D. Smith" <address@hidden>
> 
> %% Eli Zaretskii <address@hidden> writes:
> 
>   ez> --- job.c~      2006-03-20 07:03:04.000000000 +0200
>   ez> +++ job.c       2006-03-25 11:49:30.996394500 +0200
>   ez> @@ -2297,6 +2297,12 @@ construct_command_argv_internal (char *l
>   ez>                   0 };
>   ez>    char*  sh_chars;
>   ez>    char** sh_cmds;
>   ez> +#elif defined(HAVE_DOS_PATHS)
>   ez> +  /* This is required if the MSYS/Cygwin ports (which do not define
>   ez> +     WINDOWS32) are compiled with HAVE_DOS_PATHS defined, which uses
>   ez> +     sh_chars_sh[] directly (see below).  The value is identical to
>   ez> +     the one above for WINDOWS32 platforms.  */
>   ez> +  static char sh_chars_sh[] = "#;\"*?[]&|<>(){}$`^";
>   ez>  #elif defined(__riscos__)
>   ez>    static char sh_chars[] = "";
>   ez>    static char *sh_cmds[] = { 0 };
> 
> It seems uncool to have the same content twice.
> 
> Can't we break out this setting from the different places it is now and
> put it into one place, tested by HAVE_DOS_PATHS?

All the system-dependent stuff there is extremely uncool.  I
understand you want to deal with such ugliness (in job.c and
elsewhere) after 3.81 is released.  Since we are very close to the
release, I'd recommend to leave this band-aid as it is now, and
revisit it after the release, as part of removing as much of the
ifdefs as possible.