RCS file: /home/rgb/Src/CVSROOT/libwulf/libwulf.cvs.time,v Working file: libwulf.cvs.time head: 1.44 branch: locks: strict access list: symbolic names: v_1_0_1: 1.41 v1_0_0: 1.39 v0_6_2: 1.37 v0_2_1: 1.34 v0_6_1: 1.33 v0_6_0: 1.32 v0_2_0: 1.32 v0_5_9: 1.30 v0_1_7: 1.30 v0_1_6: 1.29 v0_5_8: 1.29 v_0_1_5: 1.27 t0_1_4: 1.25 v0_5_7: 1.25 t0_1_3: 1.24 v0_1_2: 1.21 v0_1_3: 1.20 p0_1_1: 1.19 v0_1_1: 1.12 v0_1_0: 1.7 v0_0_3: 1.7 t0_0_6: 1.7 v0_0_6: 1.6 c0_0_2: 1.6 v0_0_2: 1.6 v0_0_5: 1.6 start: 1.1.1.1 wulfstat-devel: 1.1.1 keyword substitution: kv total revisions: 45; selected revisions: 45 description: ---------------------------- revision 1.44 date: 2004/12/03 03:18:40; author: rgb; state: Exp; lines: +1 -1 Just making sure that the wulfware repository install is fixed and checked in here as well... ---------------------------- revision 1.43 date: 2004/12/02 16:09:34; author: rgb; state: Exp; lines: +1 -1 Is this correct? I don't get it... ---------------------------- revision 1.42 date: 2004/12/01 16:49:13; author: rgb; state: Exp; lines: +1 -1 This is a tagged (I hope) v 1.0.1 release with the following revisions: a) The rpm generated is now gpg signed and yum-distributable with checksig = 1. b) A tiny bug was fixed in the specfile -- ln -s has to be ln -sf in order to force libwulf.so to correctly link to the current version. ---------------------------- revision 1.41 date: 2004/12/01 16:35:31; author: rgb; state: Exp; lines: +1 -1 This is now a "signed" build. ---------------------------- revision 1.40 date: 2004/09/30 17:38:02; author: rgb; state: Exp; lines: +1 -1 This is the grand checkin of version 1.0.0, which is a fully functional (however "broken" internally:-) shared library version of libwulf. With libwulf installed from this rpm, wulfstat 1.0.0 builds and (apparently) runs just like it did from libwulf.a. This is going to make certain classes of development into a PITA, but with luck I can toggle a line or to in the wulf* makefiles to use the static library during development. Another great triumph is the complete automation of the makefile's "make yum" target to create a yum directory tree and its incorporation into "make installweb" to actually push the tree into additive rsync with personal and brahma wulfware repositories. When this is stuck into the other wulf* and xmlsyd makefiles, the whole thing can be distributed and maintained by anyone anywhere with yum alone. If I'm REALLY feeling frisky, I'm going to even take the tarball and wrap it into a noarch rpm that unpacks it into /tmp, builds the rpm, and installs it! Or something else that Seth would consider Dark Evil. Heh, heh, heh...:-) ---------------------------- revision 1.39 date: 2004/09/28 22:49:35; author: rgb; state: Exp; lines: +1 -1 This is a pretty much nearly working version that builds/installs shared libraries and include files. Bumped the major number to 1 in honor of this going standalone. ---------------------------- revision 1.38 date: 2004/09/28 17:31:03; author: rgb; state: Exp; lines: +1 -1 Incremental changes. We're working on making libwulf install ALL THE WAY as a development library, so in the future wulfstat and wulflogger will build alone using an already installed libwulf. We also have to figure out how to make libwulf.a into a shared library. ---------------------------- revision 1.37 date: 2004/09/16 19:09:17; author: rgb; state: Exp; lines: +1 -1 Fixes things up for kb units. ---------------------------- revision 1.36 date: 2004/09/10 17:20:05; author: rgb; state: Exp; lines: +1 -1 This now builds and installs as a standalone rpm. I'm likely missing required includes, though. I want to eventually strip the includes out of the wulfstat source and put them ALL into libwulf or wulfware-devel or the like. ---------------------------- revision 1.35 date: 2004/07/07 13:48:42; author: rgb; state: Exp; lines: +1 -1 This is a significant checkin. Fixes: a) A true bug in read_wulfhosts returned 0 when using the localhost_only option, which effectively prevented localhost from displaying. b) A secondary/related bug (I think) in the code fragment immediately following, where instead of falling back on localhost if the wulfhosts XML was bad, it would have returned 0 (effectively displaying nothing with no message). I probably still need a better messaging system and error reporting for bad xml, but this SHOULD be catchable with the -v 1 2> log trick. c) wulfstat is (harmlessly) displaying a couple of rounds of garbage when the c(mdline) option is invoked in a network that is using old xmlsysd that doesn't support the tag. I tried fixing this by zeroing the associated data structure during init, but it didn't seem to work. I'm a bit concerned that this is a symptom of a data overrun elsewhere in the code, but it seems harmless and self-correcting for now so until I get a better failure I'll have to leave it. d) Oh, and I'm implementing this inclusion of the changelog for if/when we really do split off libwulf as a separate install. In most cases the changelog will be one level up and associated with wulfstat or wulflogger, as it really isn't possible to debug the one without the other(s). As always, the only TRUE changelog is the cvs diffs back to the beginning, but this should help a bit... ---------------------------- revision 1.34 date: 2004/07/07 12:49:32; author: rgb; state: Exp; lines: +1 -1 For some reason, the changes here aren't making it to Duke... ---------------------------- revision 1.33 date: 2004/06/20 08:40:45; author: rgb; state: Exp; lines: +1 -1 I don't know where or why, but this has to go home. ---------------------------- revision 1.32 date: 2004/05/18 21:24:17; author: rgb; state: Exp; lines: +1 -1 Just checking... ---------------------------- revision 1.31 date: 2004/05/18 15:36:30; author: rgb; state: Exp; lines: +1 -1 This one never seems to get where it is supposed to go. ---------------------------- revision 1.30 date: 2004/05/17 21:51:35; author: rgb; state: Exp; lines: +1 -1 Syncing this separately to ensure it gets there. ---------------------------- revision 1.29 date: 2004/05/14 17:16:34; author: rgb; state: Exp; lines: +1 -1 This has a trivial change, but should be good now. ---------------------------- revision 1.28 date: 2004/05/14 17:03:55; author: rgb; state: Exp; lines: +1 -1 This has a tiny fix in validate_iprange but I sense that there might be another, deeper bug. I'm hoping that it doesn't -- that the crash we were seeing was just an old library issue and already fixed. I also cleaned up the Makefile a tad to make it more consistent with the project_with_library template. ---------------------------- revision 1.27 date: 2004/05/14 16:06:35; author: rgb; state: Exp; lines: +1 -1 This is slightly fixed up/cleaned up to be consistent with my new template for projects with libraries. ---------------------------- revision 1.26 date: 2004/04/16 16:43:16; author: rgb; state: Exp; lines: +1 -1 This contains two more fixes contributed by Didier Carlier; one initialization fix for the hostptr->connected flag and one allocation fix for fields used to parse stuff, both in wulfhosts_util. Neither one should hurt anything, although I'm NOT sure that the former will fix the problem I'm seeing where after a bit webwulf stops adding hosts back correctly when they are rebooted (but otherwise seems to still work). Perhaps wulflogger is crashing but not webwulf, dunno. ---------------------------- revision 1.25 date: 2004/03/12 18:09:11; author: rgb; state: Exp; lines: +1 -1 Sending a tagged revision of what MAY be a perfect snapshot home. ---------------------------- revision 1.24 date: 2004/02/23 15:17:27; author: rgb; state: Exp; lines: +1 -1 This has had a number of important changes made over the weekend, going to Duke for permanent registry. ---------------------------- revision 1.23 date: 2004/02/21 18:37:27; author: rgb; state: Exp; lines: +1 -1 This is a fairly important bugfix. I hadn't quite "finished" version.c and it was overwriting stuff in the values struct. Oops. Hope I don't have any more crap fragments like this -- it is the dark side of using block copies to get a basic template and not doing each one at a time to perfection. Which in turn is a curse of doing a major reorganization of working subroutines -- one tries to get to a "working" state quickly so as to not debug everything at once. Just means you keep finding bugs later. Sigh. ---------------------------- revision 1.22 date: 2004/02/21 18:03:18; author: rgb; state: Exp; lines: +1 -1 This has a few small fixes. The most important one is likely proper support for a localhost-only connection via the -l flag. This is primarily useful for debugging, I suppose. The second one is that net.c is fixed so that val.interface[-1] now contains TOTAL RATES for all the interfaces present. This includes, BTW, things like loopback and usb, so be warned. AFAICT, libwulf is quite functional. The notion of writing error/debugging messages to e.g. stderr is proving lovely for debugging tools interfaced with curses code, as in bash at least this channel can be redirected at the shell level. This is ALMOST enough reason to switch to bash from tcsh, as redirects are a pain in tcsh. I think it is time to check in libwulf and wulfstat and go back to wulflogger for just a second or two to make sure all is hunky-dory there wrt these changes. We are then VERY TEMPTED to make another stab at gwulfstat, using the new and VASTLY improved libwulf library, or to use wulflogger to create a poor-man's web interface. ---------------------------- revision 1.21 date: 2004/02/20 14:53:38; author: rgb; state: Exp; lines: +1 -1 This is a set of changes that permit the SAME libwulf to be used both places and write debugging output to stderr, not stdout, allowing it to be split off from an ncurses display. Works charmlike, key step. A few other mods as I run down more problems. ---------------------------- revision 1.20 date: 2004/02/19 18:12:12; author: rgb; state: Exp; lines: +1 -1 This is a tagged version, p0_1_1. It is working, afaict, and moderately decrufted. ---------------------------- revision 1.19 date: 2004/02/19 18:11:01; author: rgb; state: Exp; lines: +1 -1 This is about ready for a new tag and major checkin. It APPEARS to be close to 100% fixed wrt all the bugs I had identified previously. As in I'm starting to handle unsigned long ints better. Still not horribly well, but better. I also have cleaned up the state logic, I think so it is a lot saner. I still have to verify this (again) over in wulfstat, but at least now wulfstat has a CHANCE of working correctly once I add the toplevel fixes associated with e.g. the new improved simplified primary work cycle. ---------------------------- revision 1.18 date: 2004/02/17 16:20:18; author: rgb; state: Exp; lines: +1 -1 Sending this in as well. ---------------------------- revision 1.17 date: 2004/02/13 22:35:26; author: rgb; state: Exp; lines: +1 -1 This is going to Duke as the common version in a very tight wulflogger. I might even have to try another wulflogger rpm... ---------------------------- revision 1.16 date: 2004/02/13 19:31:05; author: rgb; state: Exp; lines: +1 -1 THIS is a very important fix to both parse and xmlsysd_util ---------------------------- revision 1.15 date: 2004/02/13 19:34:45; author: rgb; state: Exp; lines: +1 -1 Is there anythng to go home? ---------------------------- revision 1.14 date: 2004/02/13 17:55:54; author: rgb; state: Exp; lines: +1 -1 This is being fired back to TRY to FINALLY drive a stake through the heart of the +x flag on llist.h ---------------------------- revision 1.13 date: 2004/02/13 18:12:37; author: rgb; state: Exp; lines: +1 -1 This one goes home. It's broken, but not AS broken as the one home. ---------------------------- revision 1.12 date: 2004/02/12 19:09:50; author: rgb; state: Exp; lines: +1 -1 For some reason and somehow, I failed to enter these two includes into CVS. ---------------------------- revision 1.11 date: 2004/02/12 18:05:07; author: rgb; state: Exp; lines: +1 -1 This is just ensuring that it IS at Duke. ---------------------------- revision 1.10 date: 2004/02/12 18:04:45; author: rgb; state: Exp; lines: +1 -1 *** empty log message *** ---------------------------- revision 1.9 date: 2004/02/12 17:49:21; author: rgb; state: Exp; lines: +1 -1 This doesn't quite build the rpm. I think that I have to install the include files used in order to install a static library. ---------------------------- revision 1.8 date: 2004/02/12 17:12:04; author: rgb; state: Exp; lines: +1 -1 This is it, really, a near final checkin. The only things remaining to do are fix libwulf.spec and libwulf.1 ---------------------------- revision 1.7 date: 2004/02/11 19:17:51; author: rgb; state: Exp; lines: +1 -1 This needs to go to Duke as well. ---------------------------- revision 1.6 date: 2004/02/10 20:21:23; author: rgb; state: Exp; lines: +1 -1 Just checking (in) one last time. ---------------------------- revision 1.5 date: 2004/02/10 00:39:55; author: rgb; state: Exp; lines: +1 -1 This checks in LOTS of interesting changes... we are separating out all the <tag> init and update modules very consistently. Objects, sort of. ---------------------------- revision 1.4 date: 2004/02/10 00:03:23; author: rgb; state: Exp; lines: +1 -1 This is the first two components shipped out into their own init and update modules. ---------------------------- revision 1.3 date: 2004/02/09 23:26:56; author: rgb; state: Exp; lines: +1 -1 Try again... ---------------------------- revision 1.2 date: 2004/02/09 23:22:29; author: rgb; state: Exp; lines: +1 -1 This is a virgin checkin of a real attempt at making the library into a LIBRARY, stand alone... ---------------------------- revision 1.1 date: 2004/02/09 16:33:53; author: rgb; state: Exp; branches: 1.1.1; Initial revision ---------------------------- revision 1.1.1.1 date: 2004/02/09 16:33:53; author: rgb; state: Exp; lines: +0 -0 Initial standalone =============================================================================