Re: request: warning messages

Re: request: warning messages

[Jason Merrill]

g++ does not support -Wno-implicit-int, as it is a required diagnostic,
rather than a simple warning as in C.  The real solution is for fixincludes
to fix the X headers.

Jason

Re: request: warning messages

[Jeffrey A Law]

  In message < u990exc7v6.fsf@yorick.cygnus.com >you write:
  > g++ does not support -Wno-implicit-int, as it is a required diagnostic,
  > rather than a simple warning as in C.  The real solution is for fixincludes
  > to fix the X headers.
No, that's not the real solution.

We don't want to go down that road with fixincludes.

jeff

Re: request: warning messages

[Stefan Schwarzer]

>g++ does not support -Wno-implicit-int, as it is a required diagnostic,
>rather than a simple warning as in C.  The real solution is for fixincludes
>to fix the X headers.
>
>Jason

It seems that at least in the egcs-19990124 snapshot the 
implicit-int warning of the egcs-1.1.1 release got promoted to 
an error. I get this error (on solaris 2.7, but not on DUNIX 4.0) 
even without specification of '-pedantic' and even if it occurs in 
headers included as  extern "C" {}
I enclose a quote from the docs coming with that snapshot

`-pedantic'
     Issue all the warnings demanded by strict ANSI C and ISO C++;
     reject all programs that use forbidden extensions.

     Valid ANSI C and ISO C++ programs should compile properly with or
     without this option (though a rare few will require `-ansi').
     However, without this option, certain GNU extensions and
     traditional C and C++ features are supported as well.  With this
     option, they are rejected.

I hope in the interest of the survival of my project 
that in egcs releases you keep letting me compile code that uses Xlib. 
The Xlib headers on solaris (2.7) and on many other architectures
(incl DUNIX 4.0) are oldish and have this problem. And I couldn't 
see 'fixed' X11 headers lying around after the build 
in the installation directory.

I really appreciate your efforts!, but I beg you to please give 
naive developpers like me an option to choose compromise over religion. 

-- 
Stefan Schwarzer                office:  +49-(0)711-685-7606   fax: x-3658 
Uni Stuttgart, ICA 1            e-mail:          sts@ica1.uni-stuttgart.de
Pfaffenwaldring 27              pgp public key available on request/finger 
70569 Stuttgart, Germany             http://www.ica1.uni-stuttgart.de/~sts

Re: request: warning messages

[Joe Buck]

> I hope in the interest of the survival of my project 
> that in egcs releases you keep letting me compile code that uses Xlib. 

I thought that there was going to be an item in the FAQ about this, but
I don't see it.

Use -fpermissive .

> I really appreciate your efforts!, but I beg you to please give 
> naive developpers like me an option to choose compromise over religion. 

The problem with accepting bad code by default forever is that people
will write bad code forever.  -fpermissive will allow older stuff to
compile.

Re: request: warning messages

[Gerald Pfeifer]

On Fri, 29 Jan 1999, Joe Buck wrote:
>> I hope in the interest of the survival of my project 
>> that in egcs releases you keep letting me compile code that uses Xlib. 
> I thought that there was going to be an item in the FAQ about this, but
> I don't see it.

"How do I compile X11 headers with g++"
http://egcs.cygnus.com/faq.html#X11R6

Gerald

PS: If you have any suggestions on how to improve that/the FAQ, I'll be
happy to accept a patch. ;-)  Seriously, I think there's still quite some
room for improvement.
-- 
Gerald Pfeifer (Jerry)      Vienna University of Technology
pfeifer@dbai.tuwien.ac.at   http://www.dbai.tuwien.ac.at/~pfeifer/

Re: request: warning messages

[Martin v. Loewis]

> I thought that there was going to be an item in the FAQ about this, but
> I don't see it.

http://egcs.cygnus.com/faq.html#X11R6

What have you been looking for?

Regards,
Martin

Re: request: warning messages

[Joe Buck]

> > I thought that there was going to be an item in the FAQ about this, but
> > I don't see it.
> 
> http://egcs.cygnus.com/faq.html#X11R6
> 
> What have you been looking for?

It seems that I got a truncated page -- I've been getting a really
slow transfer rate to egcs.cygnus.com today.  Yes, it is there.

Re: request: warning messages

[Jeffrey A Law]

  In message < 199901292328.PAA19522@atrus.synopsys.com >you write:
  > It seems that I got a truncated page -- I've been getting a really
  > slow transfer rate to egcs.cygnus.com today.  Yes, it is there.
Yes.  BBN and/or UUNET is still having some problems.  I've gone through
periods of 75% packet loss to the Sunnyvale Cygnus network.  Unpleasant.

I'll be placing another phone call to UUNET shortly.

jeff

Re: request: warning messages

[Andris Pavenis]

On Fri, 29 Jan 1999, Stefan Schwarzer wrote:
>>g++ does not support -Wno-implicit-int, as it is a required diagnostic,
>>rather than a simple warning as in C.  The real solution is for fixincludes
>>to fix the X headers.
>>
>>Jason
>
>It seems that at least in the egcs-19990124 snapshot the 
>implicit-int warning of the egcs-1.1.1 release got promoted to 
>an error. I get this error (on solaris 2.7, but not on DUNIX 4.0) 
>even without specification of '-pedantic' and even if it occurs in 
>headers included as  extern "C" {}
>I enclose a quote from the docs coming with that snapshot
>
>`-pedantic'

As I have tested (compiled some of this month snapshots for 
i586-pc-linux-gnulibc1 and i586-pc-msdosdjgpp, latest with some
hacking , of course)  cc1plus  silently uses -Wpedantic-errors
by default (not saying that even in assembler output when -fverbose-asm
is specified). You can find it in file gcc/cp/decl.c.

Of course one can workaround by changing specs:

*cc1plus
%{!no-permissive:-fpermissive}

or something similar. But anyway I don't think secret use of -Wpedantic-errors
by default is not nice idea (maybe only to protect newbies from using
development snapshots as thay will think it broken .... )

Last snapshot I tested was egcs-990117.

>     Issue all the warnings demanded by strict ANSI C and ISO C++;
>     reject all programs that use forbidden extensions.
>
>     Valid ANSI C and ISO C++ programs should compile properly with or
>     without this option (though a rare few will require `-ansi').
>     However, without this option, certain GNU extensions and
>     traditional C and C++ features are supported as well.  With this
>     option, they are rejected.
>
>I hope in the interest of the survival of my project 
>that in egcs releases you keep letting me compile code that uses Xlib. 
>The Xlib headers on solaris (2.7) and on many other architectures
>(incl DUNIX 4.0) are oldish and have this problem. And I couldn't 
>see 'fixed' X11 headers lying around after the build 
>in the installation directory.
>
>I really appreciate your efforts!, but I beg you to please give 
>naive developpers like me an option to choose compromise over religion. 
>

I gave change of specs that fixed this problem for me

Andris

Re: request: warning messages

[Jason Merrill]

g++ does not support -Wno-implicit-int, as it is a required diagnostic,
rather than a simple warning as in C.  The real solution is for fixincludes
to fix the X headers.

Jason

Re: request: warning messages

[Jeffrey A Law]

  In message < u990exc7v6.fsf@yorick.cygnus.com >you write:
  > g++ does not support -Wno-implicit-int, as it is a required diagnostic,
  > rather than a simple warning as in C.  The real solution is for fixincludes
  > to fix the X headers.
No, that's not the real solution.

We don't want to go down that road with fixincludes.

jeff

request: warning messages

[Stefan Schwarzer]

Trying to figure out the docs in the info to egcs-1.1.1
I am trying to suppress pages of warning messages arising in some slightly 
incompatible X-headers on our system which declare functions to be 
implicitly of 'int' return type. 

Short of requesting each single warning message separately, i was
hoping that the following command line would result in more or less 
all warnings (forget about ansi stuff for the moment) MINUS the 
implicit int warnings:
  
g++ -c life.cc  -g -W -Wall -Wno-implicit-int -MMD -frepo -I/usr/dt/include -I/usr/openwin/include

However, egcs-1.1.1 (with new-abi, on solaris 2.5.1/2.7, DUNIX 3.2,
4.0), starts the output with: 
  
/usr/openwin/include/X11/Xlib.h:2063: warning: return-type of `XSetTransientForH
int' defaults to `int'
/usr/openwin/include/X11/Xlib.h:2071: warning: return-type of `XActivateScreenSa
ver' defaults to `int'
[...]

Could it be considered in future releases to respect the order of the
-W... specifications so that -Wno-implicit-int would take precedence?
Or did I just misinterprete the doc's completely?

Thanks - Stefan. 

-- 
Stefan Schwarzer                office:  +49-(0)711-685-7606   fax: x-3658 
Uni Stuttgart, ICA 1            e-mail:          sts@ica1.uni-stuttgart.de
Pfaffenwaldring 27              pgp public key available on request/finger 
70569 Stuttgart, Germany             http://www.ica1.uni-stuttgart.de/~sts

Re: request: warning messages

[N8TM]

In a message dated 1/29/99 12:40:34 PM Pacific Standard Time, sts@ica1.uni-
stuttgart.de writes:

<< It seems that at least in the egcs-19990124 snapshot the 
 implicit-int warning of the egcs-1.1.1 release got promoted to 
 an error. I get this error (on solaris 2.7, but not on DUNIX 4.0) 
 even without specification of '-pedantic' and even if it occurs in 
 headers included as  extern "C" {} >>
FWIW, I am seeing some testsuite failures on cygwin/NT, where g++ refuses to
compile test cases when the declaration is on a separate line:

int
main();