torcs-devel Mailing List for TORCS - The Open Racing Car Simulator

On 09/05/17 19:53, Bernhard Wymann wrote:
> Hi Jeremy
> 
> Using std:: here is not a good idea (because of some other compiler
> and linking issues), could you provide a fix which uses the c
> isnan?
> 
> What is a bit weird that we use the isnan in various places, what
> makes this special location for fedora (the others seem to work for
> you)?

Hi Bernhard,

I see that by coincidence Markus Koschany hit the same thing with
Debian and came up with the same solution on the same day.

G++ 6.x defaults to C++11, which requires isnan is explicitly
qualified as std::isnan. As more distributions adopt GCC 6.x, this
will become a wider issue.

The alternatives are:

1. keep the code at C++98, in which case the standard CXXFLAGS need to
have std=c++98 or std=gnu++98.

2. compile with gcc rather than g++

3. include math.h rather than cmath

So why does only this file barf. The following files in the src
directory also use isnan:

src/tools/accc/ac3dload.cpp
src/libs/learning/policy.cpp
src/modules/simu/simuv3/simu.cpp
src/modules/simu/simuv3/collide.cpp
src/modules/simu/simuv2/simu.cpp
src/modules/simu/simuv2/collide.cpp
src/drivers/human/human.cpp
src/drivers/olethros/driver.cpp
src/drivers/olethros/geometry.cpp

There seem to be two reasons.

1. Some of these are compiled with gcc, rather than g++, so will pick
up C semantics, without namespaces.

2. Some include math.h, rather than cmath, so again follow pure C
semantics.

The example I hit was the only one to both use g++ for compilation and
include from <cmath>.

This is a new code base to me, so I leave to you to decide which
approach you prefer.

If it helps, in other C++ projects personally I have been moving to
C++11, C++14 and even C++17, since the smart pointer functionality
does a lot to get rid of memory leaks.

Best wishes,


Jeremy

> 
> Kind regards
> 
> Bernhard
> 
> On 09.05.2017 18:40, Jeremy Bennett wrote: Found compiling for
> Fedora 25:
> 
>>>> $ uname -a Linux buzz 4.9.14-200.fc25.x86_64 #1 SMP Mon Mar
>>>> 13 19:26:40 UTC 2017 x86_64 x86_64 x86_64 GNU/Linux
> 
> inan needs to be qualified as std:inan. Diff:
> 
> $ cvs diff src/drivers/olethros/geometry.cpp Index:
> src/drivers/olethros/geometry.cpp 
> ===================================================================
>
> 
RCS file: /cvsroot/torcs/torcs/torcs/src/drivers/olethros/geometry.cpp,v
> retrieving revision 1.9 diff -r1.9 geometry.cpp 373c373 <
> if (isnan(r)) { ---
>>>> if (std::isnan(r)) {
> 
> ChangeLog entry:
> 
> * src/drivers/olethros/geometry.cpp (EstimateSphere): Qualify inan
> as std::inan.
> 
> Also filed as support request 12.
> 
> HTH,
> 
> 
> Jeremy
> 
> -- Tel:     +44 (1590) 610184 Cell:    +44 (7970) 676050 SkypeID:
> jeremybennett Twitter: @jeremypbennett Email:
> jer...@em... Web:     www.embecosm.com PGP key:
> 1024D/BEF58172FB4754E1 2009-03-20
>> 
>> ------------------------------------------------------------------------------
>>
>>
>> 
Check out the vibrant tech community on one of the world's most
>> engaging tech sites, Slashdot.org! http://sdm.link/slashdot 
>> _______________________________________________ Torcs-devel
>> mailing list Tor...@li... 
>> https://lists.sourceforge.net/lists/listinfo/torcs-devel
>> 
> 

-- 
Tel:     +44 (1590) 610184
Cell:    +44 (7970) 676050
SkypeID: jeremybennett
Twitter: @jeremypbennett
Email:   jer...@em...
Web:     www.embecosm.com
PGP key: 1024D/BEF58172FB4754E1 2009-03-20

Hi Jeremy

Using std:: here is not a good idea (because of some other compiler and 
linking issues), could you provide a fix which uses the c isnan?

What is a bit weird that we use the isnan in various places, what makes 
this special location for fedora (the others seem to work for you)?

Kind regards

Bernhard

On 09.05.2017 18:40, Jeremy Bennett wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> Found compiling for Fedora 25:
> 
>> $ uname -a Linux buzz 4.9.14-200.fc25.x86_64 #1 SMP Mon Mar 13
>> 19:26:40 UTC 2017 x86_64 x86_64 x86_64 GNU/Linux
> 
> inan needs to be qualified as std:inan. Diff:
> 
> $ cvs diff src/drivers/olethros/geometry.cpp
> Index: src/drivers/olethros/geometry.cpp
> ===================================================================
> RCS file: /cvsroot/torcs/torcs/torcs/src/drivers/olethros/geometry.cpp,v
> retrieving revision 1.9
> diff -r1.9 geometry.cpp
> 373c373
> <                       if (isnan(r)) {
> - ---
>> if (std::isnan(r)) {
> 
> ChangeLog entry:
> 
> 	* src/drivers/olethros/geometry.cpp (EstimateSphere):
> 	Qualify inan as std::inan.
> 
> Also filed as support request 12.
> 
> HTH,
> 
> 
> Jeremy
> 
> - -- 
> Tel:     +44 (1590) 610184
> Cell:    +44 (7970) 676050
> SkypeID: jeremybennett
> Twitter: @jeremypbennett
> Email:   jer...@em...
> Web:     www.embecosm.com
> PGP key: 1024D/BEF58172FB4754E1 2009-03-20
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v2
> 
> iEYEARECAAYFAlkR8PYACgkQvvWBcvtHVOGFeACfV7voQ5hHUnSF3hVNj4hGU/CS
> S4QAnRCfemdeFTw3Gj6tD895sCNGER0d
> =8ovI
> -----END PGP SIGNATURE-----
> 
> ------------------------------------------------------------------------------
> Check out the vibrant tech community on one of the world's most
> engaging tech sites, Slashdot.org! http://sdm.link/slashdot
> _______________________________________________
> Torcs-devel mailing list
> Tor...@li...
> https://lists.sourceforge.net/lists/listinfo/torcs-devel
> 

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Found compiling for Fedora 25:

> $ uname -a Linux buzz 4.9.14-200.fc25.x86_64 #1 SMP Mon Mar 13
> 19:26:40 UTC 2017 x86_64 x86_64 x86_64 GNU/Linux

inan needs to be qualified as std:inan. Diff:

$ cvs diff src/drivers/olethros/geometry.cpp
Index: src/drivers/olethros/geometry.cpp
===================================================================
RCS file: /cvsroot/torcs/torcs/torcs/src/drivers/olethros/geometry.cpp,v
retrieving revision 1.9
diff -r1.9 geometry.cpp
373c373
<                       if (isnan(r)) {
- ---
> if (std::isnan(r)) {

ChangeLog entry:

	* src/drivers/olethros/geometry.cpp (EstimateSphere):
	Qualify inan as std::inan.

Also filed as support request 12.

HTH,


Jeremy

- -- 
Tel:     +44 (1590) 610184
Cell:    +44 (7970) 676050
SkypeID: jeremybennett
Twitter: @jeremypbennett
Email:   jer...@em...
Web:     www.embecosm.com
PGP key: 1024D/BEF58172FB4754E1 2009-03-20
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2

iEYEARECAAYFAlkR8PYACgkQvvWBcvtHVOGFeACfV7voQ5hHUnSF3hVNj4hGU/CS
S4QAnRCfemdeFTw3Gj6tD895sCNGER0d
=8ovI
-----END PGP SIGNATURE-----

Hi all,

I have recently adopted the Debian package of Torcs and wanted to forward
some patches to you. If there is a more appropriate way to do this than
to use this mailing list, please let me know.

format-argument.patch:

Fixes a build failure since we compile Torcs with the -format-security flag for
improved security.

gcc6-isnan.patch:

Fixes a build failure because isnan is ambiguous when GCC6 is used to compile the program.

glibc-default-source.patch:

The flags  -D_SVID_SOURCE -D_BSD_SOURCE are deprecated and cause compile time warnings.
_DEFAULT_SOURCE should be used instead.

manpage-and-spelling.patch

Fixes a couple of spelling mistakes and improves the man pages a little

Regards,

Markus