The Player Project

Patches item #2646524, was opened at 2009-02-27 21:40
Message generated for change (Tracker Item Submitted) made by Item Submitter
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=433166&aid=2646524&group_id=42445

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: None
Group: None
Status: Open
Resolution: None
Priority: 5
Private: No
Submitted By: Paul Osmialowski (newchief)
Assigned to: Nobody/Anonymous (nobody)
Summary: Player SVN trunk: some more SunStudio12 compability fixes

Initial Comment:
(I've forgotten to log-in, please delete my previous anonymous post)
Recently I've compiled boost and gsl with SunStudio12 on Solaris 10 and
this uncovered some hidden errors. By the way, integrating cmake with boost
compiled using SunCC is a tricky business.

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=433166&aid=2646524&group_id=42445

Does Logitech provide any kind of API for programming these cameras?

Paul

On Fri, 27 Feb 2009, Jon Mandrell wrote:

> Is anyone working on an interface to the PTZ unit on this camera? I can use the camerauvc driver to receive images but I am unable to locate a driver for the PTZ portion. The sphere driver appears to dislike this newer version of the camera.
>
> Jon
>
>

Hi Geoff,

I've obtained todays trunk snapshot to see how things work with sunCC. It 
is much better now, however, there are four things to consider:

1. This has nothing in common with Solaris nor SunCC, but today looking at 
playerconfig.h I noticed that all things that are defined looks like:
#define HAVE_FOO 1
except one: HAVE_XDR, which looks:
#define HAVE_XDR
(no '1' at the end) why this one should differ?

2. recently applied upcbarcode fix causes new error:
[ 34%] Building CXX object 
server/libplayerdrivers/CMakeFiles/playerdrivers.dir/__/drivers/blobfinder/upcbarcode/upcbarcode.o
"/home/guest/psg22/src/player-20090217/server/drivers/blobfinder/upcbarcode/upcbarcode.cc", 
line 255: Error: In this declaration "symbols" is of an incomplete type 
"int[][2]".
"/home/guest/psg22/src/player-20090217/server/drivers/blobfinder/upcbarcode/upcbarcode.cc", 
line 257: Error: The operand "*symbols" cannot be assigned to.
2 Error(s) detected.
*** Error code 2

My proposal was and still is to fix it like this:
--- upcbarcode.cc.old   Tu feb 17 12:37:15 2009
+++ upcbarcode.cc       Tu feb 17 13:29:04 2009
@@ -252,7 +252,7 @@

    width = this->stored_data.width;
    height = this->stored_data.height;
-  int symbols[][2];
+  int (* symbols)[2];

    if ((symbols = new int[height][2]) == NULL)
    {

The "int (* symbols)[2];" declaration was taken from some old C++ book and 
seem to be still valid.

3. HAVE_IEEEFP_H definition is set properly (at least on Solaris with 
SunStudio), however the way it is used in 
server/drivers/localization/amcl/pf/pf_vector.c looks strange for me:

#elif defined (sun) && defined (HAVE_IEEEFP_H)
   #include <ieeefp.h>
#endif

Are you sure that this header comes with Solaris only? I don't know every 
OS in the world, but first answer from google says it is present at least 
in Symbian: 
http://www.symbian.com/developer/techlib/v70sdocs/doc%5Fsource/reference/cpp/libc/ieeefp%5Fh
(and the finite() function used in pf_vector.c is defined there).
I guess, "#if defined (HAVE_IEEEFP_H)" would be enough, since whenever the 
<ieeefp.h> is present, it should be included here.

(I'm curious, why are we using "#if defined (HAVE_FOO)"? Is good old 
"#ifdef" not enough?).

4. Still C++ client library cannot be compiled due to an error (however, 
currently I'm not making any kind of use of it, so I can live without it):

Scanning dependencies of target playerc++
[ 18%] Building CXX object 
client_libs/libplayerc++/CMakeFiles/playerc++.dir/playerc++.o
"/home/guest/psg22/src/player-20090217/client_libs/libplayerc++/playerc++.h", 
line 1841: Error: Unexpected type name "player_pose2d_t" encountered.
"/home/guest/psg22/src/player-20090217/client_libs/libplayerc++/playerc++.h", 
line 1841: Error: ")" expected instead of "{".
"/home/guest/psg22/src/player-20090217/client_libs/libplayerc++/playerc++.h", 
line 1841: Error: Use ";" to terminate statements.
"/home/guest/psg22/src/player-20090217/client_libs/libplayerc++/playerc++.h", 
line 1841: Error: Use ";" to terminate statements.
"/home/guest/psg22/src/player-20090217/client_libs/libplayerc++/playerc++.h", 
line 1841: Error: Unexpected ")" -- Check for matching parenthesis.
"/home/guest/psg22/src/player-20090217/client_libs/libplayerc++/playerc++.h", 
line 1841: Error: Operand expected instead of ";".
"/home/guest/psg22/src/player-20090217/client_libs/libplayerc++/playerc++.h", 
line 1846: Error: Unexpected type name "player_pose2d_t" encountered.
"/home/guest/psg22/src/player-20090217/client_libs/libplayerc++/playerc++.h", 
line 1846: Error: ")" expected instead of "{".
"/home/guest/psg22/src/player-20090217/client_libs/libplayerc++/playerc++.h", 
line 1846: Error: Use ";" to terminate statements.
"/home/guest/psg22/src/player-20090217/client_libs/libplayerc++/playerc++.h", 
line 1846: Error: Use ";" to terminate statements.
"/home/guest/psg22/src/player-20090217/client_libs/libplayerc++/playerc++.h", 
line 1846: Error: "}" expected instead of ",".
"/home/guest/psg22/src/player-20090217/client_libs/libplayerc++/playerc++.h", 
line 1846: Error: Use ";" to terminate statements.
"/home/guest/psg22/src/player-20090217/client_libs/libplayerc++/playerc++.h", 
line 1846: Error: Unexpected ")" -- Check for matching parenthesis.
"/home/guest/psg22/src/player-20090217/client_libs/libplayerc++/playerc++.h", 
line 1846: Error: Operand expected instead of ";".
"/home/guest/psg22/src/player-20090217/client_libs/libplayerc++/playerc++.h", 
line 2009: Error: Unexpected type name "player_pose3d_t" encountered.
"/home/guest/psg22/src/player-20090217/client_libs/libplayerc++/playerc++.h", 
line 2009: Error: ")" expected instead of "{".
"/home/guest/psg22/src/player-20090217/client_libs/libplayerc++/playerc++.h", 
line 2009: Error: Use ";" to terminate statements.
"/home/guest/psg22/src/player-20090217/client_libs/libplayerc++/playerc++.h", 
line 2009: Error: Use ";" to terminate statements.
"/home/guest/psg22/src/player-20090217/client_libs/libplayerc++/playerc++.h", 
line 2009: Error: Unexpected ")" -- Check for matching parenthesis.
"/home/guest/psg22/src/player-20090217/client_libs/libplayerc++/playerc++.h", 
line 2009: Error: Operand expected instead of ";".
"/home/guest/psg22/src/player-20090217/client_libs/libplayerc++/playerc++.h", 
line 2016: Error: Unexpected type name "player_pose3d_t" encountered.
"/home/guest/psg22/src/player-20090217/client_libs/libplayerc++/playerc++.h", 
line 2016: Error: ")" expected instead of "{".
"/home/guest/psg22/src/player-20090217/client_libs/libplayerc++/playerc++.h", 
line 2016: Error: Use ";" to terminate statements.
"/home/guest/psg22/src/player-20090217/client_libs/libplayerc++/playerc++.h", 
line 2016: Error: Use ";" to terminate statements.
"/home/guest/psg22/src/player-20090217/client_libs/libplayerc++/playerc++.h", 
line 2016: Error: "}" expected instead of ",".
Compilation aborted, too many Error messages.
*** Error code 1

Paul

On Tue, 17 Feb 2009, gbiggs wrote:

> Paul Osmialowski wrote:
>> 2. server/drivers/mixed/p2os/robot_params.h, this structure has char *
>> fields initialised with troublemaking string literals, so the pointer type
>> should be const char *. Previously the same issue was corrected by someone
>> else in erratics robot_params.h, so why can't it be done the same with
>> p2os?
>
> This file is generated. The files that actually need to be fixed are the
> tcl files in server/codetools/saphconv/.
>
>> 3. server/drivers/mixed/rflex/rflex.cc the joy_control static
>> property of RFLEX class cannot be declared extern. I guess only SunCC
>> complains about it, so cmake should disable this driver if detected
>> compiler is SunCC. To be honest, I can't find where this variable is
>> instanced if it is extern here!
>> or.... let's do it like that:
>>
>> #ifdef __SUNPRO_CC
>> int                      RFLEX::joy_control;
>> #else
>> extern int               RFLEX::joy_control;
>> #endif
>
> Looking at what the code does with this variable, I don't think taking
> away the extern will leave the driver in a fully-functional state. I've
> disabled it for solaris instead.
>
>
> Geoff
>
> ------------------------------------------------------------------------------
> Open Source Business Conference (OSBC), March 24-25, 2009, San Francisco, CA
> -OSBC tackles the biggest issue in open source: Open Sourcing the Enterprise
> -Strategies to boost innovation and cut costs with open source participation
> -Receive a $600 discount off the registration fee with the source code: SFAD
> http://p.sf.net/sfu/XcvMzF8H
> _______________________________________________
> Playerstage-developers mailing list
> Playerstage-developers@...
> https://lists.sourceforge.net/lists/listinfo/playerstage-developers
>

Paul Osmialowski wrote:
> Hi Geoff,
> 
> I've obtained todays trunk snapshot to see how things work with sunCC. It 
> is much better now, however, there are four things to consider:
> 
> 1. This has nothing in common with Solaris nor SunCC, but today looking at 
> playerconfig.h I noticed that all things that are defined looks like:
> #define HAVE_FOO 1
> except one: HAVE_XDR, which looks:
> #define HAVE_XDR
> (no '1' at the end) why this one should differ?

Just a typo. It's fixed now.

> 2. recently applied upcbarcode fix causes new error:

Odd, that compiled fine under my GCC.

> 3. HAVE_IEEEFP_H definition is set properly (at least on Solaris with 
> SunStudio), however the way it is used in 
> server/drivers/localization/amcl/pf/pf_vector.c looks strange for me:
> 
> #elif defined (sun) && defined (HAVE_IEEEFP_H)
>    #include <ieeefp.h>
> #endif
> 
> Are you sure that this header comes with Solaris only? I don't know every 
> OS in the world, but first answer from google says it is present at least 
> in Symbian: 

While I'm not sure symbian is one of our target operating systems, I
guess it doesn't hurt to include the header if it's there.

> (I'm curious, why are we using "#if defined (HAVE_FOO)"? Is good old 
> "#ifdef" not enough?).

The first way looks prettier, and Player must be the prettiest girl at
the ball.

> 4. Still C++ client library cannot be compiled due to an error (however, 
> currently I'm not making any kind of use of it, so I can live without it):

That's nothing, you should see what it does on Windows.

Unfortunately, I have no idea what's causing those errors other than
probably something just before the first one. Without a solaris
environment here, there's not much I can do to investigate.

Geoff

A quick update on progress with the Win32 build:

The C++ client library now compiles. I haven't tested it, and there's a
potential issue with DLL linking corrupting some data (at least with
VC++ 2005) that I haven't investigated yet, but it will probably work.

Some utilities compile. Mainly the simple ones like playerprint and
playerprop.

Command line argument passing works now.


Geoff

gbiggs wrote:
> By now, most of us probably don't have to think about the answer when
> someone asks if Player will work on Windows. Unfortunately, we're all
> going to have to do some retraining. I've just checked in a large change
> set that adds native Windows support to Player:
> 
> http://killbots.net/images/player_win32.jpg
> 
> This work is not yet complete, but it has a functional server and the C
> client libraries. A large number of drivers also compile, although I
> haven't tested them. See below for the complete list. There is a todo
> list in the root source directory, which should give an idea of what
> works and what doesn't outside of the drivers.
> 
> So far I have tested this using Visual Studio 2005 under Windows Vista
> x64 and Visual Studio 2008 under Windows XP x64. There are copious
> warnings thanks to MS's desire to deprecate standard library functions
> in favour of their own more secure versions, but they can be ignored.
> 
> Some of the changes made may have a flow-on effect causing issues on
> other platforms. I don't expect there will be many, but just in case,
> keep an eye out and let me know if any problems crop up.
> 
> I was going to write a wiki page detailing how to build and use Player
> under Windows, but the wiki is currently broken. So I'll write it in an
> email to this list soon and put it in the wiki later.
> 
> Geoff

On Feb 17, 2009, at 9:57 AM, Paul Osmialowski wrote:

> 4. Still C++ client library cannot be compiled due to an error  
> (however,
> currently I'm not making any kind of use of it, so I can live  
> without it):
>
> Scanning dependencies of target playerc++
> [ 18%] Building CXX object
> client_libs/libplayerc++/CMakeFiles/playerc++.dir/playerc++.o
> "/home/guest/psg22/src/player-20090217/client_libs/libplayerc++/ 
> playerc++.h",
> line 1841: Error: Unexpected type name "player_pose2d_t" encountered.

Here's the code the compiler is complaining about:

  void GoTo(player_pose2d_t pos)
       {GoTo(pos,(player_pose2d_t) {0,0,0}); }

Perhaps your compiler doesn't like the compact struct initialization  
and cast inside the method call.

I was about to suggest expanding that call to declare and then refer  
to an instance of the struct.  Then I updated and noticed that Geoff  
already did that, presumably as part of getting the MS compiler to  
accept it.  So you might update and try it again on Solaris.

	brian.

hi Brian,

That solved the problem. libplayerc++ finally compiled fine with no other 
errors using SunCC, also libplayerc++ examples all compiled fine. Soon I'm 
going to compile boost and I'll see how boost support for libplayerc++ 
compiles using SunCC.

Paul

On Thu, 19 Feb 2009, Brian Gerkey wrote:

>
> On Feb 17, 2009, at 9:57 AM, Paul Osmialowski wrote:
>
>> 4. Still C++ client library cannot be compiled due to an error
>> (however,
>> currently I'm not making any kind of use of it, so I can live
>> without it):
>>
>> Scanning dependencies of target playerc++
>> [ 18%] Building CXX object
>> client_libs/libplayerc++/CMakeFiles/playerc++.dir/playerc++.o
>> "/home/guest/psg22/src/player-20090217/client_libs/libplayerc++/
>> playerc++.h",
>> line 1841: Error: Unexpected type name "player_pose2d_t" encountered.
>
> Here's the code the compiler is complaining about:
>
>  void GoTo(player_pose2d_t pos)
>       {GoTo(pos,(player_pose2d_t) {0,0,0}); }
>
> Perhaps your compiler doesn't like the compact struct initialization
> and cast inside the method call.
>
> I was about to suggest expanding that call to declare and then refer
> to an instance of the struct.  Then I updated and noticed that Geoff
> already did that, presumably as part of getting the MS compiler to
> accept it.  So you might update and try it again on Solaris.
>
> 	brian.
>
>
> ------------------------------------------------------------------------------
> Open Source Business Conference (OSBC), March 24-25, 2009, San Francisco, CA
> -OSBC tackles the biggest issue in open source: Open Sourcing the Enterprise
> -Strategies to boost innovation and cut costs with open source participation
> -Receive a $600 discount off the registration fee with the source code: SFAD
> http://p.sf.net/sfu/XcvMzF8H
> _______________________________________________
> Playerstage-developers mailing list
> Playerstage-developers@...
> https://lists.sourceforge.net/lists/listinfo/playerstage-developers
>

Well that was easier than I expected. :) Yeah, I changed that the other
day when VC++ was throwing a fit about it. It was one of those "a single
error leading to a thousand more" situations... It seems that the cl
compiler's tokeniser got completely lost after that line, but kept
trying anyway.

Geoff

Brian Gerkey wrote:
> Here's the code the compiler is complaining about:
> 
>   void GoTo(player_pose2d_t pos)
>        {GoTo(pos,(player_pose2d_t) {0,0,0}); }
> 
> Perhaps your compiler doesn't like the compact struct initialization  
> and cast inside the method call.
> 
> I was about to suggest expanding that call to declare and then refer  
> to an instance of the struct.  Then I updated and noticed that Geoff  
> already did that, presumably as part of getting the MS compiler to  
> accept it.  So you might update and try it again on Solaris.

Hello Devels,

Installation of bjam and boost on Solaris uncovered only one additional 
error in libplayerc++ examples:

Scanning dependencies of target example3
[ 88%] Building CXX object 
examples/libplayerc++/CMakeFiles/example3.dir/example3.o
"/home/guest/psg22/src/player-20090219/examples/libplayerc++/example3.cc", 
line 31: Error: mem_fun is not a member of std.
1 Error(s) detected.
*** Error code 1

According to C++ reference, mem_fun() is defined in <functional>, 
therefore this issue can be easily fixed by adding:
#include <functional>
into the examples/libplayerc++/example3.cc file. This solved the problem.

Other things compiled fine.

Paul

On Thu, 19 Feb 2009, Paul Osmialowski wrote:

> hi Brian,
>
> That solved the problem. libplayerc++ finally compiled fine with no other
> errors using SunCC, also libplayerc++ examples all compiled fine. Soon I'm
> going to compile boost and I'll see how boost support for libplayerc++
> compiles using SunCC.
>
> Paul
>
> On Thu, 19 Feb 2009, Brian Gerkey wrote:
>
>>
>> On Feb 17, 2009, at 9:57 AM, Paul Osmialowski wrote:
>>
>>> 4. Still C++ client library cannot be compiled due to an error
>>> (however,
>>> currently I'm not making any kind of use of it, so I can live
>>> without it):
>>>
>>> Scanning dependencies of target playerc++
>>> [ 18%] Building CXX object
>>> client_libs/libplayerc++/CMakeFiles/playerc++.dir/playerc++.o
>>> "/home/guest/psg22/src/player-20090217/client_libs/libplayerc++/
>>> playerc++.h",
>>> line 1841: Error: Unexpected type name "player_pose2d_t" encountered.
>>
>> Here's the code the compiler is complaining about:
>>
>>  void GoTo(player_pose2d_t pos)
>>       {GoTo(pos,(player_pose2d_t) {0,0,0}); }
>>
>> Perhaps your compiler doesn't like the compact struct initialization
>> and cast inside the method call.
>>
>> I was about to suggest expanding that call to declare and then refer
>> to an instance of the struct.  Then I updated and noticed that Geoff
>> already did that, presumably as part of getting the MS compiler to
>> accept it.  So you might update and try it again on Solaris.
>>
>> 	brian.
>>
>>
>> ------------------------------------------------------------------------------
>> Open Source Business Conference (OSBC), March 24-25, 2009, San Francisco, CA
>> -OSBC tackles the biggest issue in open source: Open Sourcing the Enterprise
>> -Strategies to boost innovation and cut costs with open source participation
>> -Receive a $600 discount off the registration fee with the source code: SFAD
>> http://p.sf.net/sfu/XcvMzF8H
>> _______________________________________________
>> Playerstage-developers mailing list
>> Playerstage-developers@...
>> https://lists.sourceforge.net/lists/listinfo/playerstage-developers
>>
>
> ------------------------------------------------------------------------------
> Open Source Business Conference (OSBC), March 24-25, 2009, San Francisco, CA
> -OSBC tackles the biggest issue in open source: Open Sourcing the Enterprise
> -Strategies to boost innovation and cut costs with open source participation
> -Receive a $600 discount off the registration fee with the source code: SFAD
> http://p.sf.net/sfu/XcvMzF8H
> _______________________________________________
> Playerstage-developers mailing list
> Playerstage-developers@...
> https://lists.sourceforge.net/lists/listinfo/playerstage-developers
>

Daniele Calisi wrote:
> Do any of you have some experience in compiling Stage on Cygwin?
> 


I use Stage 2 on MinGW (not Stage 3).  I had to make various 
modifications, search the mailing list archives or look in the patch 
tracker.

Reed

Improve  your love life with generic Viagra
  http://parkerjofamu.by.ru/index.html

   

Known as transformations of water. They have all when he
saw the man stumble and pitch heavily marked the turn of
the tide, which came to its had large snakes (wound round
their upper arms) wielding the bow called pinaka. Thou residest.