This is the mail archive of the libstdc++@gcc.gnu.org mailing list for the libstdc++ project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

Re: new testsuite failure


Phil Edwards <pedwards@disaster.jaj.com> writes:

| On Sun, Mar 04, 2001 at 08:43:15AM +0100, Gabriel Dos Reis wrote:
| > Phil Edwards <pedwards@disaster.jaj.com> writes:
| > | 
| > | That text reads like an 'informative' message rather than an error.
| > 
| > which I consider is a Good Thing :-)
| 
| I meant 'informative' in the sense of "this isn't a problem in the code, but
| you might want to pay attention," like LOG_INFO in syslog(3).  I agree that
| the message is full of information, and that is definitely a Good Thing.  :-)
| 
| Well... Let's pretend this wasn't the testsuite code, and that I'm just
| some plain ordinary user who hasn't seen the recent discussion on the gcc
| list.  If the message is only for information, why does the compile fail?
| If there is an error, I don't understand what it is.  I see a lot of "for
| your information, this is what's happening" messages, and then everything
| simply stops.  Presumably there's some final *problem* that actually causes
| the failure.

One problem might be that we're using pedwarn() for things which are
just notes.  The pedwarn thingyis overloaded.  We need to clearly
identify diagnostic categories.  In my local tree I have

	enum disgnostic_kind 
	{
          DK_ERROR,
	  DK_WARNING,
	  DK_NOTE
	};

to distinguish informative diagnostics from error and warnings.  But I
need to clean up the code before inclusion on the mainline.

| > There were
| > some discussions to that effect between me and Nathan Sidwell, but
| > that is not a trivial issue.
| 
| I believe it.  Wow, I didn't know there were that many things involved.

The main problem lies in the one-line-at-a-time-as-prefix approach.
That is a fragile a fragile scheme :-(  I would pretty much prefer

	Warning: Redefinition of class 'foo' at line <xxx>.  Previous
	         definition was made in file <xxx> at line <xxx>.

that is no feasable as too many tools relies on the file:line:
prefixing scheme.  Too bad.

-- Gaby
CodeSourcery, LLC                       http://www.codesourcery.com


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]