Yesterday I downloaded the zip file of version 0.57 to start working on a port to NeXTSTEP (UNIX). Unfortunately, currently the sources are totally windoze oriented. Since a LINUX port (or any other UNIX port) is supposedly also in the making, it would be helpful to get in contact with those who are working on a unix-port of folderol to avoid double work in de-windowization of the code.
I think one should first make a platform-neutral (i.e., text-only command line) UNIX version of the code without any of the graphical elemets, i.e., without the screen saver function. just s.th. that does the work and runs in the background with low priority.
Any comments?
Helmut
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hey, the code started out as platform-independent. It only became windows-dependent when I needed some way to maintain state between runs and the registry was the natural place to do so. Malcolm has suggested a config file instead. That I think is wise. One can actually run the thing non-graphically as is by creating an instance of a GAGeneticAlgorithm as opposed to a GAGraphicGeneticAlgorithm.
The second obstacle to platform independence is HTTP PUT commands. If we had a platform independent web access library, I could remove the platform independence in about a day or so.
Scott
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I did a little code fiddling several days ago to get the non-windows parts to compile with gcc -- email me if you want diffs. I'm not planning to do anything more until it's clearer what would be worthwhile and not duplicated work.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I think this is a first step toward UNIX. Is there a standard way to put your modified sources up on source forge for the others to download? I am not very familiar with source forge's possibilities.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
sourceforge provides access to CVS for projects hosted on sourceforge. The source should be checked into cvs so that it can be checked out edited and checked back in... that will also allow you to keep a log of any changes. My question and the only reason I havn't taken time to try and make this thing work in unix, which shouldn't be hard after config is done differently is the fact that parts of the code are removed for "security purposes". I have a real hard time coding when I can't test my results. If someone could please elaborate as to why this is a security risk. If someone wanted the information contained within they could either sniff the packets or probably just strings the compiled source.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I did not try to port any of the files that use Windows libraries, just the compute engine. I can't upload them to sourceforge for you, since the repository doesn't seem to be set up, and I'm not a project member anyway. Here's a tarball: http://www.smop.com/software/folderol.tgz. It's not very different from the original -- as Oppressor said, the core parts are relatively portable. The compiler still spits out several hundred lines of warnings -- most of them harmless, but a couple I'm not sure about.
Sorry for not replying sooner.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Great! I could compile it under NeXTSTEP with only minor modifications. However, the main() function ist still pretty basic ;-) ;-) Therefore, I could not test it yet.
Would some of the project leaders please set up the source forge CVS repository so that we can actively use it and develop the code?
...and who is going to do s.th. about the rest of the code?
Later,
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Anonymous
-
2000-10-23
I did the folderol internet server access together with malcolm.
now i am working at using a configuration file, the unix side. windows is done by malcolm.
Jonas
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hehe, i'd love to set this up in CVS. The only problem is that all my tools complain whenever I try. I think there's something that the project administrators have to do so that us developers can access the FTP & CVS areas.
Anyway here's another package, a Thread abstractor class. It'll be used to abstract the folderol GeneticAlgorithm thread launching code. You can download it at: http://home.earthlink.net/~malcolmnixon/Thread.zip
Jonas and I will get the Internet one up soon.
Malcolm.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hello,
Yesterday I downloaded the zip file of version 0.57 to start working on a port to NeXTSTEP (UNIX). Unfortunately, currently the sources are totally windoze oriented. Since a LINUX port (or any other UNIX port) is supposedly also in the making, it would be helpful to get in contact with those who are working on a unix-port of folderol to avoid double work in de-windowization of the code.
I think one should first make a platform-neutral (i.e., text-only command line) UNIX version of the code without any of the graphical elemets, i.e., without the screen saver function. just s.th. that does the work and runs in the background with low priority.
Any comments?
Helmut
Hey, the code started out as platform-independent. It only became windows-dependent when I needed some way to maintain state between runs and the registry was the natural place to do so. Malcolm has suggested a config file instead. That I think is wise. One can actually run the thing non-graphically as is by creating an instance of a GAGeneticAlgorithm as opposed to a GAGraphicGeneticAlgorithm.
The second obstacle to platform independence is HTTP PUT commands. If we had a platform independent web access library, I could remove the platform independence in about a day or so.
Scott
I did a little code fiddling several days ago to get the non-windows parts to compile with gcc -- email me if you want diffs. I'm not planning to do anything more until it's clearer what would be worthwhile and not duplicated work.
How far did you get with making gcc compile it?
I think this is a first step toward UNIX. Is there a standard way to put your modified sources up on source forge for the others to download? I am not very familiar with source forge's possibilities.
sourceforge provides access to CVS for projects hosted on sourceforge. The source should be checked into cvs so that it can be checked out edited and checked back in... that will also allow you to keep a log of any changes. My question and the only reason I havn't taken time to try and make this thing work in unix, which shouldn't be hard after config is done differently is the fact that parts of the code are removed for "security purposes". I have a real hard time coding when I can't test my results. If someone could please elaborate as to why this is a security risk. If someone wanted the information contained within they could either sniff the packets or probably just strings the compiled source.
I did not try to port any of the files that use Windows libraries, just the compute engine. I can't upload them to sourceforge for you, since the repository doesn't seem to be set up, and I'm not a project member anyway. Here's a tarball: http://www.smop.com/software/folderol.tgz. It's not very different from the original -- as Oppressor said, the core parts are relatively portable. The compiler still spits out several hundred lines of warnings -- most of them harmless, but a couple I'm not sure about.
Sorry for not replying sooner.
Great! I could compile it under NeXTSTEP with only minor modifications. However, the main() function ist still pretty basic ;-) ;-) Therefore, I could not test it yet.
Would some of the project leaders please set up the source forge CVS repository so that we can actively use it and develop the code?
...and who is going to do s.th. about the rest of the code?
Later,
I did the folderol internet server access together with malcolm.
now i am working at using a configuration file, the unix side. windows is done by malcolm.
Jonas
Greetings All,
Jonas and I have got an alpha version of the Settings code. If you feel like downloading it and playing, the source can be found at: http://home.earthlink.net/~malcolmnixon/Settings.zip
I'll be putting more files up as they come out of the forge.
Malcolm.
P.S. Constructive criticisms are accepted.
Great! Compiles fine under NeXTSTEP :-)
How about setting up a real CVS thing under sourceforge?
Hehe, i'd love to set this up in CVS. The only problem is that all my tools complain whenever I try. I think there's something that the project administrators have to do so that us developers can access the FTP & CVS areas.
Anyway here's another package, a Thread abstractor class. It'll be used to abstract the folderol GeneticAlgorithm thread launching code. You can download it at:
http://home.earthlink.net/~malcolmnixon/Thread.zip
Jonas and I will get the Internet one up soon.
Malcolm.