This build includes some major changes in the query code among some other changes. Namely Mac OS X support thanks to Hyper_Eye (obviously for Odamex and Chocolate Doom, but Skulltag might work under WINE).

Changelog (Note that a few of these may have been included in the 0.4.1 release)

Zalewa

• Sorted methods in CPP files in alphabetical order.
• Fixed incorrect method documentation in configureDlg.h and configureDlg.cpp files.
• Split MainWindow constructor into smaller methods.
• Server host/createHostCommandLine methods have less parameters now. All required parameters are now passed as single HostInfo structure.
• More cleanups to server/client launch code: executable path, working dir and parameters are now described by single CommandLineInfo structure. Also some code from host()/join() methods that was the same is now executed by single Server::runExecutable() method.
• Parameter required to launch the server is now returned by one of the "argFor..." Server class virtual methods. Default implementation returns empty string while Skulltag plugin reimplements this to return "-host".
• Replaced existing flags with pictures proposed by Nece228.
• FIXED: Query menu worked incorrectly if a plugin without master server was present.
• Skulltag plugin: added a check to return an error if number of players on a server is higher than the number of maximum clients.
• Added more country flags. Credits go again to Nece228.
• FIXED: All master clients were enabled by default after the program startup disregarding the actual values set in config/query menu. (introduced in previous revision)
• More cleanups to the code: Server::join() method now returns most of the errors through a reference argument. More cleaning is required, especially to dock buddies list.
• Buddies list now emits signal with const Server* pointer when follow buddy routine is called. This signal is connected to MainWindow::runGame().
• Added several new actions to server's context menu.
• Got rid of server info dock. It's not required anymore.
• copytextdlg.h had incorrect #define at the beginning of the file.
• "Join server" and "Obtain join command line" routines are now handled by the GUI (by MainWindow class to be more specific). This removes GUI calls from within the Server class and allows better generalization of the code. This is next step (and hopefully one of the last steps) in the process of cleaning up of the host/join code.
• In host/join routines: "-file" parameter (or rather the string returned by virtual Server::argForPwadLoading()) is now put before each PWAD.
• Several QAction* pointers are now initialized with NULL. Proper checks were also added.
• Play Offline button in create server dialog (Note: this was only tested on Windows build of Skulltag).
• Added a Server::bPingIsSet field to allow plugins to determine servers' pings on their own. For example: Skulltag plugin may use this to read ping based on the time sent back by the server.
• Skulltag plugin now sets servers' pings internally.
• Added Doxygen documentation for IP2C class.
• Added new version of IP2C compacted database which contains full country names and is even smaller than version 1. Unfortunately since IP numbers are no longer sorted in ascending order in the file the program needs to use a hash table to sort the database, thus the startup process is now a bit longer (however the startup time is still insignificantly short). Compatibility with previous database version is retained.
• Full country name can now be obtained through IP2C. This is displayed in the tooltip of Server name column as "Location".
• Fixed several problems with Server name column's tooltip.
• Hopefully fixed a crash caused by Skulltag plugin.
• Increased crash resistance of Skulltag plugin.
• Increased delay between separate servers query to 2.
• Forgot to remove a debug printf from odamex plugin.
• Added even more checks to Skulltag plugin to prevent crashes.
• Wadseeker: abort() method is now a slot.
• Buttons in Wadseeker interface are now labeled more accurately.
• User may now skip a site manually in Wadseeker's dialog box.
• Added even more crash preventing checks to skulltag plugin.
• For debug purposes Skulltag plugin will now process emitUpdated(RESPONSE_BAD) signals and print the received data to stderr.
• Transformed virtual Server::readRequest() method to be more thread safe. To make sure things remain this way Server::updated signal shouldn't be emitted from within plugins anymore.
• Temporarily disabled ability of Skulltag plugin to calculate ping on it's own to see if it helps with the crashes.
• Skulltag plugin ping calculation is now re-enabled.
• Servers are now refreshed in smaller batches. This visibly reduces the refresh speed but the pings are more accurate. Perhaps an option should be provided allowing the user to choose between more accurate pings or faster speed (in form of a combo box with 3/5 relative options).
• Cleared up the servers batch indexing.
• Solved problem with Batch class accessing a private element of Server class.
• Fixed a problem with being able to register the same server several times for refreshing process which resulted in refresh buttons being gray all the time.
• Forgot to put some stuff in the previous revision.
• Split RefreshingThread::run() into smaller methods.
• Added Kazakhstan flag.
• Added framework for logger code.
• Implemented functionality for logging mechanism and changed some of the printf()s and qDebug()s to logger methods.
• Fixed: there shouldn't be any unwanted empty lines in the log box anymore.
• Extended functionality and expanded usage of Log class.
• Main::enginePlugins is now initialized during runtime, not before main() function like it was in previous revisions. This allows to use Log during plugin initialization process.
• Added more extensions to logging system.
• Fixed: Problem with #define overlapping an object in Log class.
• Wrapped CMake install instructions in if(NOT WIN32) ... endif(NOT WIN32) block as Blzut suggested.

Blzut3

• Merged the experimentalquery branch into the trunk.
• Bumped the version up to 0.5 Beta so that I don't have to do this later and svn versions identify themselves as something other than 0.4.
• Added: Chocolate Doom Plugin
• Fixed: IP2C database was not found if the current directory wasn't the program's directory.
• Fixed: It seems when registering plugins with master servers, those without masters tried to access a non-existant index in queryMenuPorts causing the program to abort.
• Added a SkulltagVersion class to the Skulltag plugin. This allows for accurate version comparisons.
• Finally got threadless querying to perform at the same speed as with threads. Still seems to results in incorrect pings.
• Fixed: inaccurate pings in the skulltag plugin.
• Updated the Huffman code.
• Move the Huffman code into the Skulltag plugin.
• Added missing files.
• Added a small Wadseeker console app as a usage example.
• Fixed compiler errors and warnings.
• Added command line parameter --updateip2c. The update process has room for improvement, but it gets the job done. Doomseeker will terminate after the update process since it is intended to be run as root on Linux systems.
• Added --help parameter which prints the available parameters to the log and terminates.
• CMake now generates an install target so `sudo make install` can be used to install Doomseeker on Linux.
• Additionally on Linux Doomseeker now searches /usr/local/share/doomseeker/engines/ for plugins. This directory can be changed with cmake since it technically points to ${CMAKE_INSTALL_PREFIX}/share/doomseeker/engines/.
• Forgot to add the desktop file to go along with the Linux installation.
• Server console (and thus RCon) now stores the last 50 commands into memory which can be accessed by pressing up and down.

Linux users must compile Doomseeker themselves. Instructions are available on the website.
Mac OS X (Universal binary 10.4 and up)
Windows

Edit: Any bugs found in this build should go in the bugs forum with [Doomseeker] in the title.

you forgot to add "blow up host computer of anyone hosting a server with a speedyshare wad download link and/or rar". other than that, very nice work!

Edit: missing an essential file. libgcc_s_dw2-1.dll.

Missing the same file.

here's the dll. install it where doomseeker is.

http://www.speedyshare.com/files/197144 ... _dw2-1.dll

I noticed speedyshare renames the dll to not have underscores, and breaks it unless you manually rename it. Not too much hassle, but for those of you who are just as lazy as I am here is a more direct link

http://users.coffeenet.org/~vulture/libgcc_s_dw2-1.dll

The need for the dll is a bug in the version of gcc blzut3 used to compile (and did not notice before he went to bed). Should be fixed soon.

thanks a bunch.

Excellent, updated my PKGBUILD on the AUR(So Arch users can install/use this easily)

I didn't compile either of the binary releases. Zalewa does the Windows binary. Any ways, he sent me an updated zip with the dll. Same location. (Technically the Windows binary is now based on r347, but since that revision just added Hyper_Eyes Mac stuff I didn't feel a new file name was needed.)

I just want to know... (I know this isn't just Doomseeker, as this happens in Doom Connector) Why when you Start a server with either Doom Seeker, or Doom Connector, it will load every wad in your autoload section of your INI File??

I Always wondered it, and that I Think I made two posts about it. Both were left unanswered. This never happens in IDE. I think it has something to do with skulltag.dll, because it used to happen in other server browsers, that I can't name off the top of my head.

Other then that.. Nice fixes/new stuff. Just that this keeps me from using doomseeker, other then its so different from IDE, and that no matter how long I wait between refreshes, it keeps saying <Refreshed too fast, try again soon> messages on some servers.

AFAIK IDE simply ignores any existing client ini file (and thus in particular the autoload section) when starting a server, while Doomseeker uses the client ini for the server.

Are you suggesting that we should alter this behavior?

I was merely explaining what's causing The_Devil_Hunter's observation. Both approaches have their merits, I'm not sure whether one is clearly better than the other.

Although dropped packets could cause this as well, perhaps increasing the value on Options->Configure->Query->Delay between tries (ms) will help? Along with maybe decreasing the number of tries.

Well, that's the thing. I don't have autoretry on. I had it on at one point, but it kept saying that refresh message. I just click Get servers, and then refresh after about 5 mins, like IDE does.

Anyhow, I guess when I host with Doomseeker, or Doom Connector, I can disable things in Autoload... Though with Doom Connector, I cannot close Skulltag, or it closes the Server down, that unless I get a Key from Zorcher, but I doubt he will ever give it to me.

This has nothing to do with with resending queries to servers that didn't respond, as one can deduce by seeing the option that is right below, stating that it will obtain new server list every period of time. The "Number of tries" and "Delay between..." spin boxes have tooltips explaining how this works.

Also why <Refreshed too soon> error happens (assuming that everything in Doomseeker works correctly, but I'm 99% sure it does ):
Server and launcher communicate on UDP protocol, so some packets are going to be missing, sometimes even on good connections. Refresh too soon may happen if this course of events occurs:

Launcher sends the query packet to the server -> server receives query packet, sends back response -> response gets lost in the void of teh Internets -> Doomseeker doesn't receive the response packet but it doesn't assume just yet that the server won't ever respond and list it as NOT RESPONDING (for this to happen server needs to ignore the query several times, which you can specify in options) -> another query packet is sent to the server -> servers sees the new query packet from the same ip, sends back "you just refreshed me!" response -> fail

There's one way to fix this permamently: increase delay between tries to 10000. But this will extend the time the whole server list is refreshed.

Yes I mean That, and well... I'm just too used to IDE.

there is one more: server shouldnt "sends back "you just refreshed me!" response" at all or at least repeat normal response several times.
Not quite understand why skulltag developers implemented this strange feature and why didnt correct it yet..

Well, I just ran it again, and it only showed One server, with the <refreshed too soon> message. and there was 10 others that had <no response> (not sure why they even show, its odd, because it does the same in IDE).

Even after waiting like before to refresh, it didn't show all servers with that message, so I guess it fixed itself. Mind you, this was BEFORE I updated to .5b.

And once ST catches up with ZDoom 2.4, it should be possible to build Mac binaries of ST.