Unnecessary thread_db message?

Unnecessary thread_db message?

[Vladimir Prus]

Over in linux-thread-db.c, there's code like this:

	void
	check_for_thread_db (void)
	{
	  /* First time through, report that libthread_db was successfuly
	     loaded.  Can't print this in in thread_db_load as, at that stage,
	     the interpreter and it's console haven't started.  */
	
	  if (!already_loaded)
	    {
		printf_unfiltered (_("Using host libthread_db library \"%s\".\n"),
				 library);


The printf is totally unconditional. When gdb is started with the --quiet option,
this line is the only thing printed. The problem is that this message means nothing
to ordinary user -- he does not know what's thread_db is, and possibly has no idea
if, say, "/lib/tls/i686/cmov/libthread_db.so.1" is the right library to load.
This message seems more like debug print -- so can we print it only when some
verbose flag is set?

- Volodya

Re: Unnecessary thread_db message?

[Jim Blandy]

Vladimir Prus <ghost at cs.msu.su> writes:
> Over in linux-thread-db.c, there's code like this:
>
> 	void
> 	check_for_thread_db (void)
> 	{
> 	  /* First time through, report that libthread_db was successfuly
> 	     loaded.  Can't print this in in thread_db_load as, at that stage,
> 	     the interpreter and it's console haven't started.  */
> 	
> 	  if (!already_loaded)
> 	    {
> 		printf_unfiltered (_("Using host libthread_db library \"%s\".\n"),
> 				 library);
>
>
> The printf is totally unconditional. When gdb is started with the --quiet option,
> this line is the only thing printed. The problem is that this message means nothing
> to ordinary user -- he does not know what's thread_db is, and possibly has no idea
> if, say, "/lib/tls/i686/cmov/libthread_db.so.1" is the right library to load.
> This message seems more like debug print -- so can we print it only when some
> verbose flag is set?

At least.  I don't think it should be printed at all --- there could
be an 'info' subcommand to show this kind of information, but as you
say, it's not something the user normally cares about.

Re: Unnecessary thread_db message?

[Michael Snyder]

On Wed, 2007-10-17 at 16:04 -0700, Jim Blandy wrote:
> Vladimir Prus <ghost at cs.msu.su> writes:
> > Over in linux-thread-db.c, there's code like this:
> >
> > 	void
> > 	check_for_thread_db (void)
> > 	{
> > 	  /* First time through, report that libthread_db was successfuly
> > 	     loaded.  Can't print this in in thread_db_load as, at that stage,
> > 	     the interpreter and it's console haven't started.  */
> > 	
> > 	  if (!already_loaded)
> > 	    {
> > 		printf_unfiltered (_("Using host libthread_db library \"%s\".\n"),
> > 				 library);
> >
> >
> > The printf is totally unconditional. When gdb is started with the --quiet option,
> > this line is the only thing printed. The problem is that this message means nothing
> > to ordinary user -- he does not know what's thread_db is, and possibly has no idea
> > if, say, "/lib/tls/i686/cmov/libthread_db.so.1" is the right library to load.
> > This message seems more like debug print -- so can we print it only when some
> > verbose flag is set?
> 
> At least.  I don't think it should be printed at all --- there could
> be an 'info' subcommand to show this kind of information, but as you
> say, it's not something the user normally cares about.

Yeah, seems like the notification-worthy event would be if it
*fails* to find or connect to the library.  So long as that case
is covered, we could maybe conditionalize this one on info_verbose.

Hey, you ask for verbose, you're gonna get verbose.  Don't see
a reason to eliminate it entirely...