realtimebattle-devel

[Realtimebattle-devel] What about a distributed server

[Benoit R. <rou...@es...>]

Imagine a client that calculate the position of his neighbor. This will
make much less calculation for one computer... 
The calculation may be done by two or three clients, so that we can
compare the received calculation and isolate/kick the modified/cheater
server (this is for security)

In the simpliest way...

                                Robot1-------------------
                                  /\                    |
                                  |                     |
                 Robot2 makes calculation for Robot1    |
                                  |                     |
   Robot2-------------------------|                     \/
     /\                                               Robot4
     |                                                  |
     |----------------------------                      |
                                 |                      |
                                Robot3<-----------------|



Client and Server would be the same application... :-)
And maybe a second application to chat and set a game, but it is not sure.
And maybe a Generale server to collapse all the general informations

---------------------------------------------------------------------------
Benoit ROUSSEAU                    benoit.rousseau at esstin.uhp-nancy.fr

[Realtimebattle-devel] Packet length

[Benoit R. <rou...@es...>]

Packet length is a big problem. We can't use them in a string because a
string refuse to include a '\0' char. I think the string length must be
independant for the netstring message and must be sent in the socket at
the last time, just before the string...


---------------------------------------------------------------------------
Benoit ROUSSEAU                    benoit.rousseau at esstin.uhp-nancy.fr

Re: [Realtimebattle-devel] cvs commitments

[Ragnar O. <rag...@st...>]

On Mon, 4 Jun 2001, Benoit ROUSSEAU wrote:

> On Sun, 3 Jun 2001, Ragnar Ouchterlony wrote:
> > > > These are the main fixes:
> > > > * Fixed a lot of things with the configure/makefile structure.
> > > >   Note: the main program should be RealTimeBattle as before (as found in
> > > >   /src/).
> > > Well, what program should it be then? The server?
> > > I thought that the "server" should be the old RealTimeBattle program.
> > > the "client" should just be the GUI and launcher. The "robot" would be the
> > > program that runs one robot and communicate with him by a pipe.
> > > Maybe the name would be more explicite if they were RTBserver RTBclient
> > > and RTBrobot, and RealTimeBattle a link to RTBserver or something like
> > > that?
> >
> > The server program should be an integral part of the RealTimeBattle and
> > not the other way around.
> So it is just a question of name.
> The main program create a ServerSocket in a time and calculate the robots
> positions (etc.) in an other time. Maybe the main() function is at a wrong
> place for you, but I think we want to do the same think :-)
> My problem is : what to do with the client and the robot program?
>
Let me think on it for some time, I want to have a good look and decide
how it should be.

> >
> > >
> > > >
> > > > * Removed some dependencies in the Network prorgams to the main program. I
> > > >   don't know what these programs need so I don't know how to fix that yet.
> > >
> > > I have a problem with this structure : there is no more libRTB.a which is
> > > needed by the Robot program for exemple (I don't know which classes it
> > > needs, but maybe the Console one... I don't know why it is needed by the
> > > Robot, but "make" said so)
> >
> > I know it is some problems there, give me a week or so to think on it (I'm
> > quite busy right now).
> I'm also quite busy at this time, and as you can see I can't do much for
> RTB this last two month. My exams are in less than one month now so I
> wouldn't do much more than now. Than holliday :-) Maybe next year in
> gotteborg and I will have much more time I hope
>

I have my exams now, and then I shall have holiday, except that I am going
to work in the summer, but I think I will be able to do some RTB:ing on
the weekends and evenings.

/Ragnar

Re: [Realtimebattle-devel] cvs commitments

[Benoit R. <rou...@es...>]

On Sun, 3 Jun 2001, Ragnar Ouchterlony wrote:
> > > These are the main fixes:
> > > * Fixed a lot of things with the configure/makefile structure.
> > >   Note: the main program should be RealTimeBattle as before (as found in
> > >   /src/).
> > Well, what program should it be then? The server?
> > I thought that the "server" should be the old RealTimeBattle program.
> > the "client" should just be the GUI and launcher. The "robot" would be the
> > program that runs one robot and communicate with him by a pipe.
> > Maybe the name would be more explicite if they were RTBserver RTBclient
> > and RTBrobot, and RealTimeBattle a link to RTBserver or something like
> > that?
> 
> The server program should be an integral part of the RealTimeBattle and
> not the other way around.
So it is just a question of name.
The main program create a ServerSocket in a time and calculate the robots
positions (etc.) in an other time. Maybe the main() function is at a wrong
place for you, but I think we want to do the same think :-)
My problem is : what to do with the client and the robot program?

> 
> >
> > >
> > > * Removed some dependencies in the Network prorgams to the main program. I
> > >   don't know what these programs need so I don't know how to fix that yet.
> >
> > I have a problem with this structure : there is no more libRTB.a which is
> > needed by the Robot program for exemple (I don't know which classes it
> > needs, but maybe the Console one... I don't know why it is needed by the
> > Robot, but "make" said so)
> 
> I know it is some problems there, give me a week or so to think on it (I'm
> quite busy right now).
I'm also quite busy at this time, and as you can see I can't do much for
RTB this last two month. My exams are in less than one month now so I
wouldn't do much more than now. Than holliday :-) Maybe next year in
gotteborg and I will have much more time I hope

Benoit

Re: [Realtimebattle-devel] cvs commitments

[Ragnar O. <rag...@st...>]

On Sun, 3 Jun 2001, Benoit ROUSSEAU wrote:

>
>
> ---------------------------------------------------------------------------
> Benoit ROUSSEAU                    benoit.rousseau at esstin.uhp-nancy.fr
>
> On Mon, 14 May 2001, Ragnar Ouchterlony wrote:
>
> > I have now updated the cvs and made it compilable, with just one
> > exception, the Network programs don't link due to some unresolved
> > dependencies in the main program.
> >
> > These are the main fixes:
> >
> > * Fixed a lot of things with the configure/makefile structure.
> >   Note: the main program should be RealTimeBattle as before (as found in
> >   /src/).
>
> Well, what program should it be then? The server?
> I thought that the "server" should be the old RealTimeBattle program.
> the "client" should just be the GUI and launcher. The "robot" would be the
> program that runs one robot and communicate with him by a pipe.
> Maybe the name would be more explicite if they were RTBserver RTBclient
> and RTBrobot, and RealTimeBattle a link to RTBserver or something like
> that?

The server program should be an integral part of the RealTimeBattle and
not the other way around.

>
> >
> > * Removed some dependencies in the Network prorgams to the main program. I
> >   don't know what these programs need so I don't know how to fix that yet.
>
> I have a problem with this structure : there is no more libRTB.a which is
> needed by the Robot program for exemple (I don't know which classes it
> needs, but maybe the Console one... I don't know why it is needed by the
> Robot, but "make" said so)

I know it is some problems there, give me a week or so to think on it (I'm
quite busy right now).

/Ragnar

Re: [Realtimebattle-devel] cvs commitments

[Benoit R. <rou...@es...>]

---------------------------------------------------------------------------
Benoit ROUSSEAU                    benoit.rousseau at esstin.uhp-nancy.fr

On Mon, 14 May 2001, Ragnar Ouchterlony wrote:

> I have now updated the cvs and made it compilable, with just one
> exception, the Network programs don't link due to some unresolved
> dependencies in the main program.
> 
> These are the main fixes:
> 
> * Fixed a lot of things with the configure/makefile structure.
>   Note: the main program should be RealTimeBattle as before (as found in
>   /src/).

Well, what program should it be then? The server?
I thought that the "server" should be the old RealTimeBattle program.
the "client" should just be the GUI and launcher. The "robot" would be the
program that runs one robot and communicate with him by a pipe.
Maybe the name would be more explicite if they were RTBserver RTBclient
and RTBrobot, and RealTimeBattle a link to RTBserver or something like
that?

> 
> * Removed some dependencies in the Network prorgams to the main program. I
>   don't know what these programs need so I don't know how to fix that yet.

I have a problem with this structure : there is no more libRTB.a which is
needed by the Robot program for exemple (I don't know which classes it
needs, but maybe the Console one... I don't know why it is needed by the
Robot, but "make" said so)

> Some questions:
> 
> roussbe:
Just call me Benoit :-)
 
>	* Doesn't the Console.{cc,h} belong to the Network code rather than
>            to the main program?
>          * What is the SimpleProcess.{cc,h} (now) doing in the
>            /src/Network/common/ directory, it isn't compiled in anywhere
>            (that I can find). Is it ok if it is removed (I know I did put
>            it there, but then I thought I would do the network code)?
>          * What does the network programs depend on in the main program?
>          * Isn't it better to put some of the SubmitPacket things in
>            Packet.cc (in /src/Network/common/) rather than having the same
>            code in many places (i.e. ServerPackets.cc ChatPackets.cc etc.)?
> 
> Erik:    * Isn't it time to release v1.0.6 soon?
> 
> /Ragnar Ouchterlony
> 
> 
> _______________________________________________
> Realtimebattle-devel mailing list
> Rea...@li...
> http://lists.sourceforge.net/lists/listinfo/realtimebattle-devel
> 

Re: [Realtimebattle-devel] Problem with CVS

[Ragnar O. <rag...@st...>]

On Sat, 2 Jun 2001, Benoit ROUSSEAU wrote:

> There are problems with CVS on SourceForge... Is it still due to the Hack
> of last (?) week ?
> I can't download last changes :-(
>

It works for me, but I had some problems some weeks ago and I had to
remove /cvsroot/RealTimeBattle/ from all CVS/Repository and I changed
/cvsroot/RealTimeBattle to /cvsroot/realtimebattle in all CVS/Root (i.e. made it
lowercase). Then it worked fine for me.

/Ragnar

[Realtimebattle-devel] Problem with CVS

[Benoit R. <rou...@es...>]

There are problems with CVS on SourceForge... Is it still due to the Hack
of last (?) week ?
I can't download last changes :-(

---------------------------------------------------------------------------
Benoit ROUSSEAU                    benoit.rousseau at esstin.uhp-nancy.fr
C++ is to programming as sex is to reproduction. Better ways might
technically exist but they're not nearly as much fun. 
                                -- Nikolai Irgens

[Realtimebattle-devel] Concurence?

[Benoit R. <rou...@es...>]

http://rars.sourceforge.net/
Maybe a couple of things to learn? (In fact I don't think so, but it was
on the danews (www.linuxfr.org)

---------------------------------------------------------------------------
Benoit ROUSSEAU                    benoit.rousseau at esstin.uhp-nancy.fr
C++ is to programming as sex is to reproduction. Better ways might
technically exist but they're not nearly as much fun. 
                                -- Nikolai Irgens

[Realtimebattle-devel] parse shell args

[Benoit R. <rou...@es...>]

To do it properly (From fr.comp.lang.c++) :

   #include <string>
   #include <vector>
   #include <iostream>

   typedef std::vector<std::string> ArgListType;

   int main(int argc, char *argv[]) {

      // store arguments into a vector
      ArgListType args(argv, argv+argc);

      // display each argument on a separate line
      ArgListType::const_iterator iter = args.begin(), iterend =
args.end();

      for (; iter != iterend; ++iter)
         std::cout << *iter << std::endl;

   }

And I don't know if we use environment variables but :

   #include <string>
   #include <map>
   #include <iostream>

   class VarEnv {

    private:

      typedef std::map<std::string, std::string> base_type;
      typedef base_type::const_iterator const_iterator;
      typedef base_type::iterator iterator;

      // stores all environment variables into an associative array
      base_type _varenv;

    public:

      // give the value of an environment variable
      std::string fetch(std::string name) {
         const_iterator iter;
         if ((iter =_varenv.find(name)) != _varenv.end())
            return iter->second;
         char *value = getenv(name.c_str());
         if (value == 0)
            return "";
         _varenv[name] = value;
         return value;
      }

      std::string operator [] (std::string s) const {
         return fetch(s);
      }
      
      // whether a environment variable is defined
      bool defined(std::string name) {
         return (_varenv.find(name) != _varenv.end() || 
                 getenv(name.c_str()) != 0);
      }

      friend std::ostream& operator << (std::ostream&, const VarEnv&);

   };

   std::ostream& operator << (std::ostream& os, const VarEnv& vars) {

      VarEnv::const_iterator iter =    vars._varenv.begin(),
                           iterend = vars._varenv.end();

      for (; iter != iterend; ++iter)
         os << iter->first << " = " << iter->second << std::endl;

      return os;

   }

   int main() {

      VarEnv vars;

      vars.fetch("HOME");
      vars.fetch("USER");
      vars.fetch("EDITOR");

      std::cout << vars;

   }

For the "SUBDIRS = . server clients", I get it from a homepage, and it has
worked on my computer so maybe it depends on the system.


---------------------------------------------------------------------------
Benoit ROUSSEAU                    benoit.rousseau at esstin.uhp-nancy.fr
C++ is to programming as sex is to reproduction. Better ways might
technically exist but they're not nearly as much fun. 
                                -- Nikolai Irgens

Re: [Realtimebattle-devel] cvs commitments

[Erik O. <er...@ma...>]

On Tue, 15 May 2001, Benoit ROUSSEAU wrote:

> > These are the main fixes:
> > * changed the directory structure of /src/Network so that the common libs
> >   now resides in a common directory, this ensures that the libs is made
> >   before the bin prorgams.
> 
> use "SUBDIRS =  .  Clients Server" 
>                ^^^
> The dot on first would make the courant directorie before entering the
> others. But it is more structured like that in fact
>

No, this doesn't work at all. It will be an infinite recursion, since it
says "Enter the current directory before the current directory" ...

It is in general a bad idea to let subdirs depend on its parent, it should
be the other way around.

/Erik

Re: [Realtimebattle-devel] Re: MOM

[Benoit R. <rou...@es...>]

> I don't think it actually is a mail-server, he only used it as an analogy.

So it is a client/server application using 'transparent' sockets ?

However, I found nothing on the web but theory. The more explicit may be
this one :
http://www.xmlblaster.org/xmlBlaster/doc/whitepaper/whitepaper.html

This second one is much more for java, and I'm not sure it uses MOM in his
examples so... :
http://www.itweb.4mg.com/

> 
> /Erik

Re: [Realtimebattle-devel] Re: MOM

[Erik O. <er...@ma...>]

On Tue, 15 May 2001, Benoit ROUSSEAU wrote:

> More questions :=20
> What happend when :
> sendmail is not installed?
> When the connected client correspond to the same login on the same serv=
er? =20
> When message from the server are checked and updated in the
> /var/spool/mail every 15 min only?
>

I don't think it actually is a mail-server, he only used it as an analogy.

/Erik
=20
> Lots of more questions until I don't know more about MOM in fact :-) So=
 I
> will not continu this list.
>=20
> -----------------------------------------------------------------------=
----
> Benoit ROUSSEAU                    benoit.rousseau at esstin.uhp-nancy.=
fr
>=20
> On Tue, 15 May 2001, Erik Ouchterlony wrote:
>=20
> >=20
> > Sorry for the late answer.
> >=20
> > On Fri, 4 May 2001, Jose e=AA Garcia Maci=F1eiras wrote:
> >=20
> > > Well, I will try to explain you about MOM.
> > >=20
> > > 	MOM is Message Oriented Middleware, it's mean that to make
> > >    a comunication to another machine you don't need to establish=20
> > >    a connection, you only send a message to a "mail server" and the
> > >    aplications will catch it and answer. The first advantage is tha=
t
> > >    you don't need make any kind a connection and protocol, you don'=
t=20
> > >    work with sockets and with nothing more. Second, you could make=20
> > >    your programs in any operative sistem, and with any language, so=
=20
> > >    you only need the inteface of the system to pick up the message=20
> > >    and leave them to post.
> > >=20
> > >          Actually you have a lot of implementations of MOM in Java,=
 C
> > >    ... well, I will try to make it, if you want to add to RTB, it's=
 you.
> > >=20
> >=20
> > It sounds like a very useful tool for rtb, so if you want to work on =
it ,
> > please go ahead.=20
> >=20
> >=20
> > Some questions:
> >=20
> > Do you have any link to an introduction of MOM?
> >=20
> > Which external tools are needed?
> >=20
> > /Erik
> >=20
> >=20
> > _______________________________________________
> > Realtimebattle-devel mailing list
> > Rea...@li...
> > http://lists.sourceforge.net/lists/listinfo/realtimebattle-devel
> >=20
>=20
>=20
> _______________________________________________
> Realtimebattle-devel mailing list
> Rea...@li...
> http://lists.sourceforge.net/lists/listinfo/realtimebattle-devel
>=20

Re: [Realtimebattle-devel] Re: MOM

[Benoit R. <rou...@es...>]

More questions :=20
What happend when :
sendmail is not installed?
When the connected client correspond to the same login on the same server? =
=20
When message from the server are checked and updated in the
/var/spool/mail every 15 min only?

Lots of more questions until I don't know more about MOM in fact :-) So I
will not continu this list.

---------------------------------------------------------------------------
Benoit ROUSSEAU                    benoit.rousseau at esstin.uhp-nancy.fr

On Tue, 15 May 2001, Erik Ouchterlony wrote:

>=20
> Sorry for the late answer.
>=20
> On Fri, 4 May 2001, Jose e=AA Garcia Maci=F1eiras wrote:
>=20
> > Well, I will try to explain you about MOM.
> >=20
> > =09MOM is Message Oriented Middleware, it's mean that to make
> >    a comunication to another machine you don't need to establish=20
> >    a connection, you only send a message to a "mail server" and the
> >    aplications will catch it and answer. The first advantage is that
> >    you don't need make any kind a connection and protocol, you don't=20
> >    work with sockets and with nothing more. Second, you could make=20
> >    your programs in any operative sistem, and with any language, so=20
> >    you only need the inteface of the system to pick up the message=20
> >    and leave them to post.
> >=20
> >          Actually you have a lot of implementations of MOM in Java, C
> >    ... well, I will try to make it, if you want to add to RTB, it's you=
=2E
> >=20
>=20
> It sounds like a very useful tool for rtb, so if you want to work on it ,
> please go ahead.=20
>=20
>=20
> Some questions:
>=20
> Do you have any link to an introduction of MOM?
>=20
> Which external tools are needed?
>=20
> /Erik
>=20
>=20
> _______________________________________________
> Realtimebattle-devel mailing list
> Rea...@li...
> http://lists.sourceforge.net/lists/listinfo/realtimebattle-devel
>=20

[Realtimebattle-devel] Re: MOM

[Erik O. <er...@ma...>]

Sorry for the late answer.

On Fri, 4 May 2001, Jose e=AA Garcia Maci=F1eiras wrote:

> Well, I will try to explain you about MOM.
>=20
> 	MOM is Message Oriented Middleware, it's mean that to make
>    a comunication to another machine you don't need to establish=20
>    a connection, you only send a message to a "mail server" and the
>    aplications will catch it and answer. The first advantage is that
>    you don't need make any kind a connection and protocol, you don't=20
>    work with sockets and with nothing more. Second, you could make=20
>    your programs in any operative sistem, and with any language, so=20
>    you only need the inteface of the system to pick up the message=20
>    and leave them to post.
>=20
>          Actually you have a lot of implementations of MOM in Java, C
>    ... well, I will try to make it, if you want to add to RTB, it's you.
>=20

It sounds like a very useful tool for rtb, so if you want to work on it ,
please go ahead.=20


Some questions:

Do you have any link to an introduction of MOM?

Which external tools are needed?

/Erik

Re: [Realtimebattle-devel] cvs commitments

[Benoit R. <rou...@es...>]

> These are the main fixes:
> * changed the directory structure of /src/Network so that the common libs
>   now resides in a common directory, this ensures that the libs is made
>   before the bin prorgams.

use "SUBDIRS =  .  Clients Server" 
               ^^^
The dot on first would make the courant directorie before entering the
others. But it is more structured like that in fact

> Some questions:
> 
> roussbe: * Doesn't the Console.{cc,h} belong to the Network code rather than
>            to the main program?
As Console would only be use with the MetaServer (Server don't have to
write anything and Clients use a GUI), I though it would be better to let
it here, but we can move it to the src directory.

>          * What is the SimpleProcess.{cc,h} (now) doing in the
>            /src/Network/common/ directory, it isn't compiled in anywhere
>            (that I can find). Is it ok if it is removed (I know I did put
>            it there, but then I thought I would do the network code)?
I forgot to remove it (Or maybe I will use it for the robot client).

>          * What does the network programs depend on in the main program?
???
What do you mean by this?

>          * Isn't it better to put some of the SubmitPacket things in
>            Packet.cc (in /src/Network/common/) rather than having the same
>            code in many places (i.e. ServerPackets.cc ChatPackets.cc etc.)?
There are a lot of things like that to do I think. I will have a look at
it.


Benoit

[Realtimebattle-devel] cvs commitments

[Ragnar O. <rag...@st...>]

I have now updated the cvs and made it compilable, with just one
exception, the Network programs don't link due to some unresolved
dependencies in the main program.

These are the main fixes:

* changed the directory structure of /src/Network so that the common libs
  now resides in a common directory, this ensures that the libs is made
  before the bin prorgams.

* Some smaller warnings has been fixed in the network code.

* SubmitPacket::SubmitPacket(int,int, ...)
  has been changed to five constructors similar to
  SubmitPacket::SubmitPacket(int,string&)
  with up to five strings possible.
  This is to get away from sending string through ... (which isn't
  possible).

* Made sure nothing comes after #endif directives, which makes a lot of
  warning messages with the gcc-2.96 (as distributed by redhat).
  This change has also been applied to the v1.0.x branch.

* Fixed a lot of things with the configure/makefile structure.
  Note: the main program should be RealTimeBattle as before (as found in
  /src/).

* Removed some dependencies in the Network prorgams to the main program. I
  don't know what these programs need so I don't know how to fix that yet.

Some questions:

roussbe: * Doesn't the Console.{cc,h} belong to the Network code rather than
           to the main program?
         * What is the SimpleProcess.{cc,h} (now) doing in the
           /src/Network/common/ directory, it isn't compiled in anywhere
           (that I can find). Is it ok if it is removed (I know I did put
           it there, but then I thought I would do the network code)?
         * What does the network programs depend on in the main program?
         * Isn't it better to put some of the SubmitPacket things in
           Packet.cc (in /src/Network/common/) rather than having the same
           code in many places (i.e. ServerPackets.cc ChatPackets.cc etc.)?

Erik:    * Isn't it time to release v1.0.6 soon?

/Ragnar Ouchterlony

Re: [Realtimebattle-devel] Network code

[Benoit R. <rou...@es...>]

> I have been looking into the Network code and I am trying to make it
> compile and so forth, I have made some changes that I haven't commited yet
> (especially to the Makefile.am and configure-scripts).
All right, I wait for the commit :-) In fact, I don't really know how to
continue. There are too many things to change at the same time. I think we
should write a road map or something like that.


> 
> The main RTB program should still be the RealTimeBattle program in the
> /src dir, if you need some things from the main program I can consider
> making a common lib to be included in the Network apps.
Would it be the client or the server?

I saw an other game (Teg) where the client can launch the server than
connect to it. I thing it's kind a good idea, don't you?



Benoit

[Realtimebattle-devel] Network code

[Ragnar O. <rag...@st...>]

I have been looking into the Network code and I am trying to make it
compile and so forth, I have made some changes that I haven't commited yet
(especially to the Makefile.am and configure-scripts).

The main RTB program should still be the RealTimeBattle program in the
/src dir, if you need some things from the main program I can consider
making a common lib to be included in the Network apps.

/Ragnar

Re: [Realtimebattle-devel] Event Handle

[Benoit R. <rou...@es...>]

On Wed, 25 Apr 2001, Erik Ouchterlony wrote:
> 
> 
> On Tue, 24 Apr 2001, Benoit ROUSSEAU wrote:
> 
> > All right, this will be the last message of the day :-)
> > 
> > For the previous message, I don't know what it was. This wasn't working at
> > school, but I has worked at home ... don't understand...
> > 
> > Than I'd tryed to link it use a EventHandler. 
> > For this metaserver, I don't think it will be possible because metaserver
> > would have to run for a while without being disconnected. Maybe we will
> > have problem with long time execution.
> > For the server, I don't know how to organize the work. Should we split the
> > Event.cc source code in ServerEvent.cc ClientsEvent.cc ... like for 
> > sockets or packets, or should everything stay in the same file?
> > 
> 
> Well, I don't know how you will connect the metaserver with rtb, so it
> depends on that. But I still think it is better to put the different
> EventClass in the file that uses it, not collect them all in Event?T.cc.

It was my opinion too :-)

> Isn't it better to let the metaserver be independent of rtb? The server
> and client parts are more important to build into rtb, so it is easy to
> start a network game.

Until now, the metaserver use the same communication process than the
server or the clients, using packets. And the structure of the
metaserver socket is also similar to the server one. I think it is better
to put it in the same directory so we don't have to duplicate the code.

Re: [Realtimebattle-devel] Robot list

[Erik O. <er...@ma...>]

On Wed, 25 Apr 2001, Benoit ROUSSEAU wrote:

> where do you think we are going to put the robot and the arena lists when
> we set the tournament? 
> ie : a chater submit one of his robots. What to do with it?
> Courantly, I have a tempory structure :
> struct tmp_robot {
> 	string name;  //The path of the robot
> 	NetConnection *chater; //The chater who has sent the robot
> };
> 
> and a tempory list :
> list<tmp_robot*> tmp_robots;
> 
> both are in the socketserver declaration but I don't think this is the
> best way.
>  
> I can add a robot to a list, and delete it when the chater leave the game.
> I can say the chater to execute the Robot application. This application
> connect correctly to the server and will be able to run a real robot very
> soon.
> 

Maybe we could use the InfoClasses and InformationDistributor for this
too, not only for the guis.

An idea: Why not let the InformationDistributor act as a 'mailman', through
which any messages between different parts of rtb are sent. Just add a
'to' and a 'from' field to the message and the InformationDistributor will
deliver.

/Erik

Re: [Realtimebattle-devel] Event Handle

[Erik O. <er...@ma...>]

On Tue, 24 Apr 2001, Benoit ROUSSEAU wrote:

> All right, this will be the last message of the day :-)
> 
> For the previous message, I don't know what it was. This wasn't working at
> school, but I has worked at home ... don't understand...
> 
> Than I'd tryed to link it use a EventHandler. 
> For this metaserver, I don't think it will be possible because metaserver
> would have to run for a while without being disconnected. Maybe we will
> have problem with long time execution.
> For the server, I don't know how to organize the work. Should we split the
> Event.cc source code in ServerEvent.cc ClientsEvent.cc ... like for 
> sockets or packets, or should everything stay in the same file?
> 

Well, I don't know how you will connect the metaserver with rtb, so it
depends on that. But I still think it is better to put the different
EventClass in the file that uses it, not collect them all in Event?T.cc.

Isn't it better to let the metaserver be independent of rtb? The server
and client parts are more important to build into rtb, so it is easy to
start a network game.

/Erik

Re: [Realtimebattle-devel] InfoClasses.h

[Erik O. <er...@ma...>]

On Tue, 24 Apr 2001, Benoit ROUSSEAU wrote:

> There seems to be a problem in object_arc_pos_info with the 'double or'
> argument of the constructor. Have you ever notice a problem like that? 
> Changing the name in 'outer_r' seems to compile better??? (maybe 'or' is a
> keyword, but I never eared anything like that...  

If it causes problems anywhere, we should change it. Not so good name
anyway ;)

/Erik

Re: [Realtimebattle-devel] Makefile.am

[Ragnar O. <rag...@st...>]

On Tue, 24 Apr 2001, Benoit ROUSSEAU wrote:

> > I looked at the changes the made in configure.in and Makefile.am.
> >
> > Please don't break the code for others when you make the changes:
> >
> >  AC_LIBTOOL_DLOPEN
> >  AC_DISABLE_STATIC
> >  AM_PROG_LIBTOOL
> >  AM_GNU_GETTEXT
> >
> > These tools are really needed in v2.0!
> So what are the tools corresponding to this declaration? where to find
> them (in which 'rpm' I mean, because if he don't have them that mean
> that I have forget one I think...)
>

The packages you need is the following, or a later version if you find one!

libtool-1.3.5
gettext-0.10.35

/Ragnar

[Realtimebattle-devel] Robot list

[Benoit R. <rou...@es...>]

where do you think we are going to put the robot and the arena lists when
we set the tournament? 
ie : a chater submit one of his robots. What to do with it?
Courantly, I have a tempory structure :
struct tmp_robot {
	string name;  //The path of the robot
	NetConnection *chater; //The chater who has sent the robot
};

and a tempory list :
list<tmp_robot*> tmp_robots;

both are in the socketserver declaration but I don't think this is the
best way.
 
I can add a robot to a list, and delete it when the chater leave the game.
I can say the chater to execute the Robot application. This application
connect correctly to the server and will be able to run a real robot very
soon.

---------------------------------------------------------------------------
Benoit ROUSSEAU                    benoit.rousseau at esstin.uhp-nancy.fr