This file contains the most recent changes to the COW client, with example .xtrekrc options to deal with them, and the clips from the documentation that explain them. If you're upgrading from an old version of COW, this file should make it a painless experience. ***************************************************************** 1) Change Log ***************************************************************** 3.00 pl2 Jan. 7, 2000 - New default portSwap for UDP_PORTSWAP feature [siegl] - System wide .xtrekrc file set to "/etc/netrekrc" [siegl] - Cambot recording replay [koconnor] - various Win32 fixes [siegl] - V,B core dump with rabit ears fix [siegl] - remove DEFINES (BRMH NEW_DASHBOARD_2 RABBIT_EARS) [siegl] - MAX_MACRO increased to 255 [siegl] - -B options for automatic bug sumition [siegl] - new sample netrekrc file [siegl] - WindowMaker docking support [siegl] - Win32 Sound support [siegl] 3.00 pl1 Mar. 27, 1999 - Added UDP queries to the metaserver. [cameron, villalpando] - Added Multiple metaserver support. [cameron, villalpando] - Metaserver KEYGOD alias: clientkeys@clientkeys.netrek.org [villalpando] - New xtrekrc variables: metaverbose, metaUDPcache, metaType [cameron, villalpando] - modified behavior of metaserver to do multiple metaservers [cameron] - changed metaserver menu. Added age of data and refresh button [cameron] - new option, -M for TCP metaserver [cameron] - changed option -m for UDP metaserver [cameron] - Port to Cygnus Win32 [siegl] - Autoconfig detects Cygnus Win32 [siegl] - Win32 fixes [Gosselin] - Fix Pixmap transparancy [siegl] - Added hockey rink background XPM on galactic [siegl] 3.00 pl0 Feb. 28, 1998 - full color client [kellen] - metaserver defaults to: metaserver.netrek.org [sven] - new cow mailing list alias: cow@netrek.org [sven] - 32 views color pixmaps [Shawn] -- recording feature [koconnor] - KDE sound support (.wav files) [siegl] - Tab key fixes [koconnor] - OpenVMS fixes [cameron] - Auto upgrade via netscape [siegl] - whydead message fixes [siegl] - GMP 2 key generator fix [siegl] ***************************************************************** 2) Cut and past changes to make COW work like it used to. ***************************************************************** 3.00 pl0 The "-b" option disables the use of color pixmaps by the client. ***************************************************************** 3) Cut and past changes to turn on *all* new features. ***************************************************************** 3.00 pl1 WARNING: Sometimes the colored ships have color disturbences when having the shields up. In this case start COW with the option: [-n] use nearest colors in shared colormap 3.00 pl0 All new .xtrekrc entrys with defaults are: # backgroundPix: 1 # cloakPix: 1 # explosionPix: 1 # fedPix: 1 # indPix: 1 # kliPix: 1 # mapPix: 1 # oriPix: 1 # romPix: 1 # shipPix: 1 # weaponPix: 1 **************************************************************** 4) New manual sections **************************************************************** 3.00 pl0 3.1.8 NO PIXMAPS OPTION The "-b" option disables the use of color pixmaps by the client. 3.2 Pixmaps (Full Color COW) With the release of COW 3.00, dynamic color images are available. No color images have been compiled into the client, so without the additional files (described below) the client will behave as before. 3.2.1 Setup The xpm files should be available at the same site from which you got the client, in a file named pixmaps.tgz (PIXMAPS.ZIP for windows users). >>> YOU MUST DOWNLOAD AND UNPACK THIS FILE TO USE THE COLOR FEATURES. <<< It should create a subdirectory named "pixmaps" which should have several (obviously named) subdirectories. There should be several XPM files in each (except for Planets, which has a further subdirectory). UNIX users will see that they are gzipped to save space. You do NOT need to ungzip them unless you do not have gzip on your machine. >>> DO NOT REARRANGE OR RENAME THESE FILES IF YOU WANT TO USE THEM. <<< You need to add a line to your .xtrekrc telling the client where to look for the pixmaps. If you do not, it will assume that they are in a subdirectory of the directory you are in when you start the client. The option is called "pixmapDir". Tilde and environment variables WILL NOT WORK. Relative paths will only work if you always start netrek from the same directory. You should be ready to rock and roll. Fire it up. You may see some warnings about not being able to read some pixmaps. Some of the pixmaps that the client looks for haven't been drawn yet. Feel free to make your own set. OTOH, if you see any lines which read "TYPE PIXMAPS NOT AVAILABLE" it means that none of a certain type of pixmap were found. Check to make sure that the pixmaps are where you told it to look. If they are, and you are on a UNIX system, you may not have gzip installed. Go get it from any GNU mirror and either install it or use it to ungzip the XPM files. 3.2.2 Configuration In addition to simply creating your own XPMs with a paint program, you may want greater control over the pictures used. For example, you may find the explosions are too pretty, and you are dying because you forgot to dodge. The crude approach is to just remove that pixmap. The client will default back to the standard bitmaps in this case. The more elegant approach is to turn off just those pixmaps you don't like and keep the rest. This also allows you to switch back and forth WITHOUT having to exit and restart. So if the machine you are playing on is busy today, you can turn off the pixmaps until things improve, then switch back to full color without losing your 5 kills. Pixmaps can be turned on or off in groups on the new "Pixmap Menu" in the options window (shift-O). Each line in the window also corresponds to an .xtrekrc resource which you can use to set the initial values. If one type of pixmaps is not available, you will be unable to turn on that option. resource name default description indPix on \ fedPix on | Control whether or not the XPMs romPix on | for the ships of a given team kliPix on | should be used oriPix on / weaponPix on Torps and plasmatorps & their clouds explosionPix on ship and starbase explosions cloakPix on fade-in/-out and cloak icon mapPix on Color Planet icons on galactic (replaces the "colorgalactic" option) backgroundPix on Background stars & genocide/gb images (replaces the "babes" option) ownerhalo off Draws a colored ring around each planet on the galactic as a convenience, the option "shipPix" may be used to control all of the ship XPMs in one line. ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ + POWER USERS + + + + The MegaResource "pixFlags" can be used in your .xtrekrc + + to save a bit of typing. Simply bitwise OR together the + + things you want turned off: + + + + 0x0001 IND pixmaps + + 0x0002 FED pixmaps + + 0x0004 ROM pixmaps + + 0x0008 KLI pixmaps + + 0x0010 ORI pixmaps + + + + 0x0020 Weapons + + 0x0040 Explosions + + 0x0080 Cloaking + + 0x0100 Galactic Map Planet Icons + + + + 0x0400 Backgrounds + + + + 0x1000 Halos + + + + so, for exaple, no halos and no explosions would be + + specified as: (0x1040=4160) + + + + pixFlags: 4160 + + + + + + (Note that this OVERRIDES all the other resources) + ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ If you hate them all, you can either set the "pixmapDir" to "None" or start the client with the -b (bitmap only) command line option. Then go ahead and delete all of the XPMS. Go ahead. We don't mind at all. It's not like we put any WORK into this ... :,-( 3.2.3 Babes/M31 and Generalized Backgrounds Gone. You can put up any picture you like when you GENO, GB, or just enter or hit shift-K. Just specify the genocide.xpm, ghostbust.xpm and/or hello.xpm. It's really none of my business what you look at in your off time. Absolutely no picture will be shown if you do not have an XPM in the specified place. It didn't belong in the client in the first place. DEAL WITH IT. And BTW, the ' key (quote) has the default action of retiling your local and galactic windows with the normal background (either black or your specified pixmap) to repair the damage done by the other possible pix. 3.2.4 AGRI pixmaps and FEATURE_PACKETS The client shows a different pixmap for AGRI planets than all others. This was announced, voted on and overwhelmingly adopted (80%+ in favor). But JUST IN CASE, this option can be disabled at the server by use of the feature packet "AGRI_PIXMAP". Users can choose to remove the AGRI.xpm file. The client will default to using the regular planet pixmap if it is missing. **************************************************************** 5) Changes details **************************************************************** 3.00 pl0 From: James Cameron Subject: COW 2.02pl2 OpenVMS Patch [Resend] The last time I built COW for OpenVMS was back in 1994, when it was 1.01pl0. Since my department is releasing a CD-ROM to it's private customers I was asked to provide a variety of Netrek software, as pre-built binaries for OpenVMS, Digital UNIX, and others. So I took 2.02pl2 and got it working again on OpenVMS. The new DEC C compiler on OpenVMS was quite reasonable about the code and had no significant trouble. I submit the patch to you for consideration. The changes were - removal of fopen declarations from functions, as it conflicted with the OpenVMS variant of (const char *, const char *, ... ) - a macro replacement of the select() call so that OpenVMS specific code could be called; the X-Windows implementation on OpenVMS does not support the use of ConnectionNumber() and select() to wait on X events. Code to support this was split out into a new file, vmsio.c. - suppression of the tools window behaviour, in the same manner as on WIN32, since SIGCHLD and popen() are only available in recent OpenVMS versions. - minor changes to vmsutils.c to comply with ANSI standard on initialisation of structures, (before I found the qualifier to turn off strict ANSI checking). - a new file config.h.vms, with OpenVMS version checking to work around old versions of OpenVMS. I've attached the patch to this message in MIME format. Strangely, the patch included cflags.c changes. I'm not sure why cflags.c should be in the manifest ... it created a needless diff output for it. Would you have preferred me moving the OpenVMS specific files into a subdirectory? For your interest, the 1.01pl0 port I did evolved into a bridged client, using serial I/O that was OpenVMS specific. This allowed me to run the Netrek protocol over a 14k4 modem without any SLIP or PPP overhead. Now, of course, OpenVMS has joined the ranks of operating systems providing serial internet protocol layers, and so my work is no longer required. Moving to 33k6 also helped. ;-) -- James Cameron (cameron@stl.dec.com) Digital Equipment Corporation (Australia) Pty. Ltd. A.C.N. 000 446 800 3.00 pl1 From: Jonathan Ellis (1) Replaced WINSIDE #define for SMALL_SCREEN with TWINSIDE and GWINSIDE for independent specification of tactical and galactic sizes on machines with painfully small displays. (I prefer to use a smaller galactic with a full-size tactical.) A lot of other functions used WINSIDE besides the actual drawing routines--mostly computing default window geometries. I tried to keep it consistent with the original look, but I might have goofed in a couple places. (2) cowmain changed so that if no server is specified, it tries to connect to the metaserver. This is much better behavior IMHO. Note: COW.DOC states, "The compiled in default server is bronco.ece.cmu.edu" This seems to be outdated: grep bronco *.c *.h defs.h:#define DEFAULT_SERVER "bronco.ece.cmu.edu" grep DEFAULT_SERVER *.c *.h defaults.c: * Otherwise we try DEFAULT_SERVER. */ defs.h:#define DEFAULT_SERVER "bronco.ece.cmu.edu" Note that the only time DEFAULT_SERVER was used is inside a comment (that, apparently, doesn't reflect the actual code. (3) added description of highlightFriendlyPhasers, censorMessages, and the new default server behavior to COW.DOC. (4) removed from COW.DOC: "Multiline macros are no-longer allowed at all. Primarily this is based on network traffic arguments." I suppose it's historical, but outdated documentation isn't cool. (5) Added note to COW.DOC that not specifying any plist options will default to old plist. (This wasn't obvious to me because 2 is called the "traditional" plist... perhaps an unintended connotation.) (6) removed from socket.c #ifdef FOR_MORONS { /* it _is_ an 'info' borg, * after all. ;-) */ extern int For_Morons; if (For_Morons) return; } #endif FOR_MORONS and For_Morons appeared nowhere else. I assume this is obsolete code. (7) removed from data.h extern int babes; after all, they're gone. ;) (8) merged the "Moo" docs (section 8 of COW-DOC) with the rest of it. It had to happen sometime. Cope. (This includes a bunch of minor changes that I won't ennumerate.) a. replaced an inaccurate line of defwin.c with "warnShields", BOOL_DEF, "Shield color based on alert status", b. someone who understands the server's ntserv/socket.c needs to explain what udpClientRecv really does. "simple" and "fat" don't say a whole lot. I don't have time to try to learn the server-side code right now. (9) the option menu as well as the example .xtrekrc claimed that the udpDebug resource took different options: 0 = OFF 1 = ON (conect msgs only) 2 = ON (verbose output) but in the code, udpDebug == 1 behaved exactly the same as udpDebug == 2 I tried to changed this to reflect the documentation by taking out all the UDPDIAG but printUdpInfo from udpDebug == 1. (10) (11) defaults.c: deleted legacy code from findDefaults dealing with server- specific .xtrekrc files. The code was broken (try it: move .xtrekrc to .xtrekrc-continuum, or whatever--it won't work). I don't know how long it's been broken, but if (serverName) { sprintf(serverfile, NETREKRCSERVER, serverName); if (findfile(serverfile, file)) return 1; sprintf(serverfile, XTREKRCSERVER, serverName); if (findfile(serverfile, file)) return 1; } never gets executed because serverName hasn't been initialized by the (first) call to findDefaults. If anybody actually used this feature there would have been complaints that it was broken... since, apparently, nobody really cared, I just got rid of it. Simplify, simplify. ;) Also: reread_defaults in input.c and newinput.c didn't work the way it was apparently intended to if no rc file was present. In order to get it to issue the warning that none was found, some things had to be changed, including initializing defaultsFile in data.c to "" instead of ".xtrekrc" (12) removed unused variables from dmessage.c char cmupigcall[80]; char pigcall[160]; (13) The ghost start option is still present in the code, but it doesn't seem to work. I have no clue what's wrong (possibly a server problem?) or if the idea is just inherently problematic. I left the documentation in, though, and even added it to what -u prints out. (14) Added -i to printUsage (15) Removed from options.c char *mapupdates[] = {"Don't update galactic map", "Update galactic map rarely", "Update galactic map frequently", ""}; apparently this isn't used anymore. (16) changed hard-coded geometries to be more sane, with focus on creating #ifdef-d alternates for SMALL_SCREEN. Parents of some windows were changed, so it should look pretty much like you'd expect without any foo.geometry: in your .xtrekrc. (17) Combined newPlanetBitmaps and whichNewPlanetBitmaps resources with showLocal/showGalactic. Cleaned up code in local.c. map.c doesn't perform the same things local.c does with equivalent options, but this is the bitmaps' fault. moobitmaps.h and rabbitbitmaps.h would need to be modified extensively. That's not my department. ;) ObBugReport: cloaking doesn't display correctly on my linux machine with pixmaps on; there's no state in between fully cloaked and fully uncloaked. One update you're visible, the next you're not. Very disconcerting. Other ideas: (sort of an addenum to the TODO ;) html-ize the new COW.DOC. cow.html was more than two years out of date even before my changes. ;) If COW.DOC were consistant in its use of section and subsection numbering this could be automated. a. "highlight" default value for xtrekrc options Make the xtrekrc sections of COW.DOC consistant with the shift-O menus and defwin.c (i.e. use the same categories for all three) ******************************************************************** March 1999 ******************************************************************** (1) ------ Added UDP metaserver, and multiple metaserver queries to client. Multiple metaservers is available only in UDP mode. Multiple metaservers may be listed in the metaservers variable, or a host name may have multiple IPs. Client essentially sends a '?' using UDP to the metaserver(s) and waits for a response. Client should expect anywhere from 0 to infinite responses. Response comes back in the form: r,NN\n Where r is the literal 'r', and NN is the nuber of servers returned. Server format is 1 per line in the following format, comma separated: (from the metaserver code snippet from disp_udp.c) sprintf(bp,"%s,%d,%d,%d,%d,%d,%c\n", sp->hostname, /* host name of server */ sp->port, /* port number of server */ extendedstatus, /* metaserver status code */ now - sp->last_update, /* age of data in seconds */ sp->player_count, /* count of players on server */ sp->queue_size, /* length of wait queue */ sp->type[0] ); /* type code, B, P, etc */ Two notes are required: 1) Line length is defind in the metaserver as: MAX_HOSTNAME+1 +6+1 +3+1 +6+1 +3+1 +3+1 +1+1 +1 host port stat age play que type \n where MAX_HOSTNAME is #define MAX_HOSTNAME 64 An example output is: r,19 hp06.ee.ualberta.ca,2592,6,244,0,0,P megalag.netrek.org,2592,4,1780,0,0,F moore.machine2.dsl.psn.net,2592,6,1804,0,0,B netrek.unh.edu,2592,6,524,0,0,B paradise.games.uk.demon.net,2592,6,1604,0,0,P tanya.ucsd.edu,2592,4,1781,0,0,P defiant.theo-physik.uni-kiel.de,2592,3,630,0,0,P europa.informatik.uni-frankfurt.de,2592,3,676,0,0,P mit.netrek.org,2592,3,676,0,0,B netrek.cs.mcgill.ca,2592,3,796,0,0,B netrek.syd.att.net.au,2592,3,800,0,0,B se.netrek.org,2592,3,864,0,0,B soda.csua.berkeley.edu,2592,3,878,0,0,B spamburger.openface.ca,2592,3,871,0,0,B kirk.hal-pc.org,2592,2,121,1,0,B continuum.us.netrek.org,2592,2,123,6,0,B hockey.netrek.org,2592,2,123,6,0,H monster.wormwood.org,2592,2,203,7,0,C guinness.crhc.uiuc.edu,2592,1,323,0,4,B 2) In TCP mode, if a game server is dead, the metaserver will print out sp->why_dead instead of sp->status. Since metastatuslevel filters out the metaserver output based on the game server status, sp->status had to be expanded in the metaserver to extended status when reporting back to the client. From disp_udp.c in the metaserver: /* this if structure is to satisfy COW's server filtering based on */ /* the TCP ouput of the metaserver. TCP was text, this is only #s */ if (sp->status == SS_NOCONN ) extendedstatus = (sp->why_dead == WD_TIMEOUT) ? 6 : 4; else extendedstatus = sp->status; And in the client, parsemeta.c: #define SS_WORKING 0 #define SS_QUEUE 1 #define SS_OPEN 2 #define SS_EMPTY 3 #define SS_NOCONN 4 #define SS_INIT 5 /* not a real metaserver number, but overcomes a limitation of dropping */ /* text description of sp->why_dead */ #define SS_TOUT 6 This note is here mainly for people looking at metaserver and client code at the same time. (2) --- A refresh button was added to the client selection window to re-query the metaservers. (3) --- KEYGOD changed to clientkeys@clientkeys.netrek.org (4) --- New command line option behavior was added. -m and -M set the default connection type to the metaserver UNLESS metaType is defined in the .xtrekrc. -m Default to UDP -M Default to TCP Out of bounds values for metaType default to TCP.