Re: [PATCH 3/4] GDBServer: introduce --server-stderr command line option

[Eli Zaretskii <eliz@gnu.org>]

> From: Cleber Rosa <crosa@redhat.com>
> Cc: crosa@redhat.com, areis@redhat.com
> Date: Fri, 20 Mar 2015 23:34:24 -0300
> 
> This command line option will redirect all of the gdbserver's own
> output (always sent to stderr) to a separate file. This feature
> makes it possible to distinguish between the inferior process stderr
> and gdbserver's own stderr.
> [...]
> +@cindex @option{--server-stderr}, @code{gdbserver} option
> +The @option{--server-stderr} option tells @code{gdbserver} that any content
> +that it would normally generate itself to its own @code{stderr} should be
> +redirected to another file. This is useful if you want to keep the
> +inferior @code{stderr} separate from the one generated by @code{gdbserver}.

Note how the text description of the rationale for the change you
posted to explain it to us is much more useful than the technical
description in the manual.  Why not say in the manual what you told
us?

From the user POV, the fact that gdbserver uses stderr for its own
output is an implementation detail that is almost unimportant.  (It
could be important for redirection purposes, but the command-line
option you introduce eliminates the need to redirect in most cases,
right?)  What _is_ important is that gdbserver's own output will be
redirected to that file, and that important information gets lost in
the confusing "it would normally generate itself to its own 'stderr'",
which leaves unanswered the question what part of gdbserver's output
is "normally" excluded from that.

For this reason, I suggest to name the option "--server-output".

And I think we want a NEWS entry for this change.

> +@item --server-stderr
> +Instruct @code{gdbserver} to redirect its own @code{stderr} to another
> +file.

The option requires an argument, so the argument should be mentioned
with the option and referenced in the text that describes it.

> +static int
> +set_server_stderr (char *path)
> +{
> +  FILE *tmp;
> +
> +  tmp = fopen (path, "w");
> +  if (tmp == NULL)
> +    {
> +      fprintf (stderr,
> +	       "Could not open server stderr file '%s': %s\n",
> +	       path, strerror(errno));
> +      return -1;
> +    }
> +  server_stderr = tmp;
> +  return 0;
> +}
> +
>  void
>  monitor_show_help (void)
>  {
> @@ -3017,6 +3036,7 @@ gdbserver_usage (FILE *stream)
>  	   "                            none\n"
>  	   "                            timestamp\n"
>  	   "  --remote-debug        Enable remote protocol debugging output.\n"
> +	   "  --server-stderr=PATH  Redirect server's STDERR to file at PATH.\n"
>  	   "  --version             Display version information and exit.\n"
>  	   "  --wrapper WRAPPER --  Run WRAPPER to start new programs.\n"
>  	   "  --once                Exit after the first connection has "
> @@ -3186,7 +3206,13 @@ captured_main (int argc, char *argv[])
>  
>    while (*next_arg != NULL && **next_arg == '-')
>      {
> -      if (strcmp (*next_arg, "--version") == 0)
> +      if (strncmp (*next_arg, "--server-stderr=",
> +		   sizeof ("--server-stderr=") - 1) == 0)
> +	{
> +	  char *path = *next_arg + (sizeof ("--server-stderr=") - 1);
> +	  set_server_stderr (path);
> +	}
> +      else if (strcmp (*next_arg, "--version") == 0)

AFAIK, GNU Coding Standards frown on using "path" for anything that is
not PATH-style list of directories.  So please use "file" or "file
name" here.

Thanks.