| 00:03.42 | *** join/#bzflag forrestv` (n=forrestv@c-66-229-251-213.hsd1.fl.comcast.net) |
| 00:03.45 | *** join/#bzflag oo7_ (n=Alsadair@71.237.185.51) |
| 00:04.46 | JeffM | hmmm 120k seems large for a list server. |
| 00:05.12 | JeffM | http server I mean |
| 00:37.28 | *** join/#bzflag Agent86 (n=chris@gentoo/user/Agent86) |
| 00:48.54 | *** join/#bzflag bryjen (n=bryjen@2002:4cb1:bee3:0:0:0:0:1) |
| 00:48.54 | *** mode/#bzflag [+v bryjen] by ChanServ |
| 00:49.47 | *** join/#bzflag Macrosoft (n=Macrosof@c-75-65-81-245.hsd1.la.comcast.net) |
| 00:50.23 | Macrosoft | code freeze? |
| 00:53.07 | Winny | no, it actually says coda freezing. |
| 00:53.09 | *** join/#bzflag rob1n (n=emp@unaffiliated/rob1n) |
| 01:04.29 | Macrosoft | w00t! AMD released specs on GPU's |
| 01:05.11 | Winny | ...woot... old |
| 01:05.45 | Macrosoft | now just need people to make driver |
| 01:05.52 | TD-Linux | :P |
| 01:05.58 | TD-Linux | and I already bought a nvidia :( |
| 01:06.02 | TD-Linux | i nfact |
| 01:06.06 | Macrosoft | i have both |
| 01:06.07 | TD-Linux | it arrived 2 hours ago |
| 01:06.32 | Macrosoft | x1550 and geForce6200OC |
| 01:07.18 | Macrosoft | now i can have the elusive 5 monitor desktop |
| 01:07.20 | Macrosoft | :p |
| 01:17.56 | *** join/#bzflag Doctor-Who (n=Net_Smar@adsl-068-209-251-067.sip.asm.bellsouth.net) |
| 01:24.28 | *** join/#bzflag bier_ (n=bier@p54A538A7.dip0.t-ipconnect.de) |
| 01:54.06 | *** join/#bzflag ViciousM1 (n=max@adsl-065-013-159-245.sip.rdu.bellsouth.net) |
| 01:54.20 | blast007 | JeffM and brlcad: this may be of interest to you: http://www.ibm.com/developerworks/web/library/wa-ltwebserv/ (Lightweight Web Servers) |
| 02:04.55 | *** join/#bzflag Theme97 (n=Theme97@about/essy/phpwriter/Theme97) |
| 02:12.13 | *** join/#bzflag rob1n (n=emp@unaffiliated/rob1n) |
| 02:21.56 | JeffM | neat, thanks blast007 |
| 02:22.42 | blast007 | JeffM: something else that's useful is 'ab' (apache benchmark) |
| 02:23.00 | blast007 | usefull for hitting a web server with a lot of traffic and seeing how long the requests take |
| 02:23.07 | JeffM | cool |
| 02:23.53 | CIA-25 | BZFlag: 03bthansen * r15882 10/trunk/bzflag/plugins/fastmap/fastmap.cpp: Fix typo in log message |
| 02:56.26 | *** join/#bzflag iwcylabumf (n=chatzill@user-0cev208.cable.mindspring.com) |
| 03:04.15 | *** join/#bzflag BenUrban (n=benurban@c-68-49-10-243.hsd1.md.comcast.net) |
| 03:08.10 | *** join/#bzflag tankzilla (n=alex@ool-4572e362.dyn.optonline.net) |
| 03:09.02 | tankzilla | hey guys, how can i set the time to stay so i don't have to push (-) to make it brighter? |
| 03:18.40 | *** join/#bzflag Legoguy_ (n=Legoguy@adsl-71-143-234-117.dsl.emhril.sbcglobal.net) |
| 03:19.33 | *** join/#bzflag tuxd00d (n=tuxinato@128.187.129.156) |
| 03:23.56 | *** join/#bzflag JeffM (n=JeffM@unaffiliated/jeffm) |
| 03:33.11 | *** join/#bzflag tankzilla (n=alex@ool-4572e362.dyn.optonline.net) |
| 03:33.24 | *** part/#bzflag tankzilla (n=alex@ool-4572e362.dyn.optonline.net) |
| 04:11.26 | brlcad | blast007: that is interesting, thanks |
| 04:12.04 | brlcad | i'm thinking this would be something really good to test for the list server, given our hit rate |
| 04:12.36 | brlcad | I still think we're simply mysql-bound at the moment by far, but apache isn't far behind if we get off phpbb's tables |
| 04:12.56 | *** join/#bzflag daxxar_ (i=daxxar@daxxar.com) |
| 04:15.59 | brlcad | howdy daxxar_ |
| 04:23.04 | AAA_awright | MSVC doesn't do MySQL... I wonder why... |
| 04:29.57 | *** join/#bzflag user__ (i=4b84007a@gateway/web/cgi-irc/zeebrothers.net/x-f6991b4e32e84151) |
| 05:34.04 | *** join/#bzflag L4m3r (n=l4m3r@about/essy/warning/L4m3r) |
| 05:34.04 | *** mode/#bzflag [+v L4m3r] by ChanServ |
| 06:38.48 | *** join/#bzflag lepoulpe303 (n=LePoulpe@about/essy/MobileTarget/LePoulpe303) |
| 07:22.41 | *** join/#bzflag CIA-24 (n=CIA@208.69.182.149) |
| 07:30.38 | *** join/#bzflag brlcad (n=sean@bz.bzflag.bz) |
| 08:21.19 | *** join/#bzflag cosmol (n=dave@71.86.121.200) |
| 08:26.09 | *** join/#bzflag Legoguy_ (n=Legoguy@adsl-70-131-34-122.dsl.emhril.sbcglobal.net) |
| 08:52.37 | *** join/#bzflag Legoguy (n=Legoguy@adsl-71-143-226-186.dsl.emhril.sbcglobal.net) |
| 09:08.11 | *** join/#bzflag Legoguy_ (n=Legoguy@adsl-71-143-224-153.dsl.emhril.sbcglobal.net) |
| 09:17.03 | *** join/#bzflag Legoguy (n=Legoguy@adsl-71-143-230-101.dsl.emhril.sbcglobal.net) |
| 09:32.20 | *** join/#bzflag Legoguy_ (n=Legoguy@adsl-71-143-228-161.dsl.emhril.sbcglobal.net) |
| 10:14.52 | *** join/#bzflag Legoguy (n=Legoguy@adsl-70-131-36-99.dsl.emhril.sbcglobal.net) |
| 10:18.27 | *** join/#bzflag Suspect (n=Suspect@68-114-71-106.dhcp.plt.ny.charter.com) |
| 10:30.37 | *** join/#bzflag CBG` (n=CBG@about/essy/fan/CBG) |
| 10:33.26 | daxxar | Howdy brlcad :-) |
| 10:39.28 | Manu | hi daxxar :) |
| 10:41.36 | daxxar | Hi Manu =) |
| 10:42.31 | *** join/#bzflag Legoguy_ (n=Legoguy@adsl-71-143-227-124.dsl.emhril.sbcglobal.net) |
| 10:57.17 | *** join/#bzflag Legoguy (n=Legoguy@adsl-70-131-35-12.dsl.emhril.sbcglobal.net) |
| 11:01.29 | *** join/#bzflag CIA-24 (n=CIA@208.69.182.149) |
| 11:08.22 | *** join/#bzflag Quol (n=Ian@69.90.0.146) |
| 11:13.58 | *** join/#bzflag Pommes_ (n=Pommes@about/essy/frenchy/Pommes) |
| 11:17.27 | *** join/#bzflag Legoguy (n=Legoguy@adsl-70-131-65-225.dsl.emhril.sbcglobal.net) |
| 11:24.32 | *** join/#bzflag Legoguy_ (n=Legoguy@adsl-71-143-237-198.dsl.emhril.sbcglobal.net) |
| 11:25.29 | *** join/#bzflag tupone (n=atupone@pdpc/supporter/active/Tupone) |
| 11:25.29 | *** mode/#bzflag [+v tupone] by ChanServ |
| 11:36.02 | *** part/#bzflag Macrosoft (n=Macrosof@c-75-65-81-245.hsd1.la.comcast.net) |
| 11:41.27 | *** join/#bzflag Legoguy (n=Legoguy@adsl-71-143-233-212.dsl.emhril.sbcglobal.net) |
| 11:55.13 | *** join/#bzflag Legoguy_ (n=Legoguy@adsl-70-131-66-182.dsl.emhril.sbcglobal.net) |
| 12:21.12 | *** join/#bzflag whodaman- (n=whodaman@p54A3EA1D.dip.t-dialin.net) |
| 12:27.42 | *** join/#bzflag Quol (n=Ian@69.90.0.146) |
| 12:28.09 | Quol | anyone here to test something for me? I am at work and so can't fire up bzflag :( |
| 12:37.42 | Quol | oh well - nvm, I will test later....thanks |
| 12:41.19 | *** join/#bzflag KTL (n=KTL@213.219.188.29.adsl.dyn.edpnet.net) |
| 12:43.01 | Epyon | ~svn |
| 12:43.02 | ibot | Subversion (aka SVN) is version control software that aims to be a better CVS than CVS. See http://subversion.tigris.org/. |
| 12:43.07 | Epyon | un |
| 12:43.10 | Epyon | ~bzsvn |
| 12:43.11 | ibot | hmm... bzsvn is http://my.bzflag.org/w/BZFlag_SVN and svn co https://bzflag.svn.sourceforge.net/svnroot/bzflag/trunk/bzflag bzflag |
| 12:47.04 | *** join/#bzflag Legoguy (n=Legoguy@adsl-70-131-44-102.dsl.emhril.sbcglobal.net) |
| 13:05.03 | *** join/#bzflag Apathy (n=Matt@74.86.45.130) |
| 13:33.42 | *** join/#bzflag Legoguy (n=Legoguy@adsl-71-143-233-126.dsl.emhril.sbcglobal.net) |
| 14:02.39 | *** join/#bzflag whodaman- (n=whodaman@p54A3EA1D.dip.t-dialin.net) |
| 14:09.25 | *** join/#bzflag Legoguy_ (n=Legoguy@adsl-70-131-71-162.dsl.emhril.sbcglobal.net) |
| 14:53.04 | *** part/#bzflag lepoulpe303 (n=LePoulpe@about/essy/MobileTarget/LePoulpe303) |
| 14:59.01 | *** join/#bzflag Legoguy_ (n=Legoguy@70.142.202.128) |
| 15:08.00 | *** join/#bzflag Legoguy (n=Legoguy@adsl-70-131-75-183.dsl.emhril.sbcglobal.net) |
| 15:12.44 | *** join/#bzflag gsnedders (n=gsnedder@host86-139-120-102.range86-139.btcentralplus.com) |
| 15:21.37 | *** join/#bzflag Legoguy_ (n=Legoguy@70.142.202.215) |
| 15:29.38 | *** join/#bzflag Legoguy_ (n=Legoguy@adsl-71-143-226-10.dsl.emhril.sbcglobal.net) |
| 15:29.53 | *** join/#bzflag JeffM (n=JeffM@unaffiliated/jeffm) |
| 15:30.41 | *** join/#bzflag Tedius (n=Tedius@musictech.MUSC.CWRU.Edu) |
| 15:33.19 | Tedius | There is no way to specify surface normals, only vertex normals, correct? |
| 15:36.21 | Tedius | Does that mean that normals in bzw meshes are really only used for smoothing? |
| 15:44.18 | Tedius | and... when tristripping one =must= use smoothing normals? |
| 15:45.37 | Tedius | Therefore, one cannot use LOD's without calculating vertex normals? |
| 15:50.35 | JeffM | yes you need vertex normals |
| 15:51.01 | JeffM | strips work by sharing vertex data, so all the data at the vert must be the same, XYZ, UV, and IJK |
| 15:51.33 | JeffM | ether that, or you break the strip at that point |
| 15:52.01 | whodaman- | how do you set up CPU prioritys on a mac? |
| 15:52.25 | JeffM | a box would be one of the worst case shapes for doing tristrips |
| 15:52.34 | JeffM | it's best for organic shapes |
| 15:53.29 | Tedius | I see, |
| 15:54.04 | Tedius | The vectors for these normals are direction only, right? |
| 15:54.08 | JeffM | yes |
| 15:54.17 | JeffM | that's why they are vectors ;) |
| 15:54.40 | Tedius | and the x y z are in relation to the world? |
| 15:55.04 | Tedius | so normal 0 0 1 would be for a top face, right? |
| 15:55.48 | JeffM | they are in the same Coordinate system as the verts |
| 15:55.58 | JeffM | so yes, baring any transformation z is up |
| 15:56.22 | Tedius | ok, its not in relation to the face |
| 15:56.38 | JeffM | no |
| 15:59.29 | JeffM | it's just normal 3d modeling stuff |
| 15:59.55 | blast007 | (since CIA isn't responding) just fixed bz_setMaxWaitTime so that the value can be increased as well, and changed the plugin_HTTP to use a 2 second maxWaitTime when there are no connections |
| 16:00.08 | JeffM | cool |
| 16:00.17 | blast007 | JeffM: I'm working on the webstats plugin a bit too ;) |
| 16:00.32 | JeffM | I was thinking of making bzfs keep one wait time per plug-in then have bzfs just use the lowest one |
| 16:00.32 | Tedius | Thanks for the help Jeff, you guys do great work |
| 16:00.36 | JeffM | so that we don't blow stuff out |
| 16:00.47 | blast007 | and now that I've made bzfs not use 100% of the CPU, I can test a little easier |
| 16:01.11 | JeffM | Tedius, the speed dif for doing strips and fans is not as much as just doing LOD |
| 16:01.46 | JeffM | heh |
| 16:01.47 | Tedius | oh, I see |
| 16:02.08 | JeffM | LOD is a huge speedup cus it just flat out LOWERS polygon counts |
| 16:02.30 | Tedius | but if it is just a cube, there would be no reason to, right? |
| 16:02.39 | JeffM | strips and fans can speed up the drawing of more then one triangle. but only in specific cases |
| 16:02.55 | JeffM | yeah if it's a cube, it's already doing the minimum stuff |
| 16:03.29 | Tedius | I like strips and fans because they're easier to code |
| 16:03.31 | JeffM | it woudl be "possible" to take it from the 12 faces down to 2, if you got super tricky |
| 16:03.34 | Tedius | (less typing) |
| 16:03.41 | JeffM | but it'd look like crap |
| 16:03.55 | JeffM | for any complex mesh your going to want to use a modeler |
| 16:04.34 | CIA-24 | BZFlag: 03blast007 * r15883 10/trunk/bzflag/ (4 files in 2 dirs): |
| 16:04.38 | CIA-24 | BZFlag: Fix bz_setMaxWaitTime so we can increase the wait time too |
| 16:04.43 | CIA-24 | BZFlag: Use a 2 second maxWaitTime for HTTP when we have no connections |
| 16:04.50 | blast007 | heh, about time |
| 16:05.53 | Tedius | for complex meshes, it is too complicated to use LOD's |
| 16:06.08 | JeffM | not really |
| 16:06.09 | Tedius | unless I create a script for it |
| 16:06.11 | Tedius | no? |
| 16:06.22 | JeffM | many modelers have polygon reduction tools |
| 16:06.27 | JeffM | export more then one version |
| 16:06.32 | JeffM | at difrent reses |
| 16:06.38 | JeffM | then make those into the LOD levels |
| 16:06.53 | Tedius | oooh, nifty |
| 16:12.20 | Tedius | So, I will make the mesh like I want it with all the normals and texcoords I want, export it to bzw, create the low res version, create a script to convert the format to drawInfo format... |
| 16:12.30 | Tedius | and copy and paste that into the LOD level |
| 16:12.52 | Tedius | copy and paste the tris and quads information |
| 16:17.32 | Tedius | hmm, will the poly reduction maintain the same vertex index numbers? |
| 16:19.35 | *** join/#bzflag gsnedders (n=gsnedder@host86-139-120-102.range86-139.btcentralplus.com) |
| 16:21.24 | *** join/#bzflag cosmol (n=dave@71.86.121.200) |
| 16:23.23 | *** join/#bzflag Legoguy (n=Legoguy@adsl-70-131-74-117.dsl.emhril.sbcglobal.net) |
| 16:36.29 | CIA-24 | BZFlag: 03blast007 * r15884 10/trunk/bzflag/plugins/webstats/ (statTemplates.cpp statTemplates.h webstats.cpp): |
| 16:36.29 | CIA-24 | BZFlag: Use a table for the player list |
| 16:36.29 | CIA-24 | BZFlag: Page is now valid HTML 4.01 Strict |
| 16:41.18 | *** join/#bzflag L4m3r (n=l4m3r@about/essy/warning/L4m3r) |
| 16:41.18 | *** mode/#bzflag [+v L4m3r] by ChanServ |
| 16:46.31 | *** join/#bzflag Erroneous (n=dtremena@67-131-219-2.dia.static.qwest.net) |
| 16:50.52 | *** join/#bzflag gsnedders (n=gsnedder@host86-139-120-102.range86-139.btcentralplus.com) |
| 16:56.47 | *** join/#bzflag Legoguy_ (n=Legoguy@70.142.196.11) |
| 16:59.53 | *** join/#bzflag JeffM (n=JeffM@unaffiliated/jeffm) |
| 17:04.22 | Thumper_ | blast007: I don't think that fix will work |
| 17:04.37 | blast007 | which fix? |
| 17:04.44 | Thumper_ | the set max wait time fix |
| 17:04.53 | Thumper_ | other plugins are depending on whatever they set as the max |
| 17:04.58 | Thumper_ | and you're clobbering it probably |
| 17:05.11 | Thumper_ | so if you load two plugins and one says max wait is 0.5s |
| 17:05.20 | Thumper_ | but the next plugin you load says max wait is 2s |
| 17:05.31 | Thumper_ | then you'll wait 2 s which is too long for the first plugin to function properly |
| 17:05.32 | Thumper_ | (I think) |
| 17:05.40 | blast007 | yup |
| 17:06.08 | blast007 | 09:58:35 < JeffM> I was thinking of making bzfs keep one wait time per plug-in then have bzfs just use the lowest one |
| 17:06.12 | Thumper_ | but it's still better than a 100% busy cpu loop :) |
| 17:06.32 | Thumper_ | ah I missed that. |
| 17:13.13 | *** join/#bzflag bryjen (n=bryjen@63.147.94.149) |
| 17:13.13 | *** mode/#bzflag [+v bryjen] by ChanServ |
| 17:13.14 | CIA-24 | BZFlag: 03blast007 * r15885 10/trunk/bzflag/plugins/webstats/ (statTemplates.cpp statTemplates.h webstats.cpp): Add CSS to color the table for each team. |
| 17:34.57 | *** join/#bzflag gsnedders (n=gsnedder@host86-139-120-102.range86-139.btcentralplus.com) |
| 17:48.40 | *** join/#bzflag Legoguy (n=Legoguy@adsl-70-142-193-97.dsl.emhril.sbcglobal.net) |
| 18:09.38 | *** join/#bzflag Pimpi (n=frank@gondolin.pimpi.org) |
| 18:11.31 | *** join/#bzflag SpazzyMcGee (n=Spazzy_M@client-86-31-53-128.nrth.adsl.virgin.net) |
| 18:14.54 | JeffM | Thumper_, blast007 yeah I'll try to get that done when I can |
| 18:14.57 | JeffM | today has been hell |
| 18:15.19 | Thumper_ | JeffM: thx. I'm away this weekend so can't do anything useful before next week |
| 18:17.15 | JeffM | the unfortonate problem is I don't know what plug-in is calling the command, so I think I'll have to do a named wait time |
| 18:17.22 | JeffM | then to be kind, you register one with your wait time. |
| 18:17.27 | JeffM | with a name |
| 18:17.34 | JeffM | and pull it when you don't care anymore |
| 18:19.07 | Thumper_ | can't you just use the address of the variable as the identifier - if each plugin has their own distinct value the address is unique |
| 18:20.32 | Thumper_ | but then you're forced to set the local var and call the function with that instead of a constant which is kinda sucky |
| 18:20.36 | Thumper_ | nevermind :) |
| 18:28.51 | JeffM | you mean pass it in by address, not value |
| 18:28.54 | JeffM | that'd be possible |
| 18:29.12 | Thumper_ | yeah except that has a problem with how you call it which isn't nice |
| 18:29.14 | JeffM | but it'd mean they'd have to keep a static var for it |
| 18:29.17 | JeffM | yeah |
| 18:29.21 | JeffM | a name is simpler |
| 18:29.35 | Thumper_ | if you change plugins to be a subclass of bz_Plugins (new) instead of bz_EventHandler |
| 18:29.44 | Thumper_ | and you can stash a var in there and just use it's address |
| 18:29.48 | Thumper_ | each plugin should be unique that way |
| 18:29.57 | Thumper_ | and the plugins don't need to know how the internals work |
| 18:30.07 | JeffM | that'd mean reqriting them all |
| 18:30.15 | JeffM | and makeing them into a class |
| 18:30.22 | JeffM | not just 3 exported functions |
| 18:30.41 | Thumper_ | ok... |
| 18:30.54 | JeffM | it's all dooable, but I think we are too far into this method |
| 18:31.18 | JeffM | and the benifit for a small wait state thing is minimal |
| 18:31.29 | JeffM | best would be to expose a real timer thread |
| 18:31.55 | JeffM | and just say "hey I need to be called his often, and here is my event handler" then spawn a thread on the server that just does that |
| 18:31.56 | Thumper_ | ok |
| 18:32.07 | JeffM | then its independent of the main bz loop |
| 18:32.35 | JeffM | looking back, doing the export from the plug-in as a class would have been smarter, true |
| 18:32.48 | JeffM | little more tricly to export, but still |
| 18:44.01 | *** join/#bzflag eTangenT (n=tangent@inge068164.halls.colostate.edu) |
| 18:45.34 | *** join/#bzflag Suspect (n=Suspect@68-114-71-106.dhcp.plt.ny.charter.com) |
| 19:03.21 | CIA-24 | BZFlag: 03JeffM2501 * r15886 10/trunk/bzflag/ (10 files in 8 dirs): |
| 19:03.21 | CIA-24 | BZFlag: use an optional name when setting the max wait times. let the API specify and clear those names. |
| 19:03.23 | CIA-24 | BZFlag: change all the plug-ins that call wait time functions with names. |
| 19:05.14 | forrestv | bzfs on arm segfaults, i'm researching why |
| 19:19.08 | *** join/#bzflag Epyon_ (i=Epyon@host99.180.219.81.magma-net.pl) |
| 19:33.30 | *** join/#bzflag tupone (n=atupone@pdpc/supporter/active/Tupone) |
| 19:33.30 | *** mode/#bzflag [+v tupone] by ChanServ |
| 19:51.40 | *** join/#bzflag Legoguy_ (n=Legoguy@70.142.204.131) |
| 20:02.56 | *** join/#bzflag Constitution (n=const@unaffiliated/macsforme) |
| 20:07.28 | JeffM | hey blast007 did you make it include a CSS file? or just CSS code at the top? |
| 20:07.45 | *** join/#bzflag Legoguy (n=Legoguy@adsl-71-143-234-176.dsl.emhril.sbcglobal.net) |
| 20:08.19 | blast007 | JeffM: CSS code at the top |
| 20:10.30 | brlcad | is it up somewhere? |
| 20:10.54 | blast007 | I'll throw it up on my server quick |
| 20:10.56 | blast007 | just have to build |
| 20:11.24 | JeffM | blast007, I had some images for it as well, hosted on .bz in images/webstats |
| 20:11.32 | JeffM | icon for the players section, etc.. |
| 20:11.41 | blast007 | ah, ok |
| 20:11.55 | JeffM | hadn't thought of a design or antyhing |
| 20:12.08 | JeffM | I just wanted to toss a background image behind it and give it the BZ icon at the top, etc.. |
| 20:12.28 | JeffM | maybe get saturos to make us a cool logo for it one day |
| 20:21.36 | *** join/#bzflag cosmol (n=dave@71.86.121.200) |
| 20:21.56 | JeffM | brlcad, you there? |
| 20:27.43 | brlcad | mebbie |
| 20:28.02 | brlcad | hows the o2? |
| 20:28.22 | JeffM | have to figure out how to partition the drives that Erroneous gave me |
| 20:28.45 | JeffM | been too busy writing on book to do that when I get home tho :/ |
| 20:29.11 | JeffM | Erroneous said that you had profiled the list server, and knew where it spent most of it's time, do you have that data? |
| 20:36.43 | *** join/#bzflag TimRiker (n=timr@72.11.78.6) |
| 20:36.44 | *** mode/#bzflag [+o TimRiker] by ChanServ |
| 20:40.26 | ts | blast007: Did you retest bzfs with a newer build? |
| 20:40.45 | blast007 | not yet, build failed |
| 20:41.12 | ts | :( |
| 20:43.19 | blast007 | one down, now another one ;) |
| 20:46.33 | CIA-24 | BZFlag: 03t-s * r15887 10/branches/v2_0branch/bzflag/src/bzfs/bzfs.cxx: We are not counting packets, either. We count % :) |
| 20:49.08 | *** join/#bzflag TD-Linux (n=TD-Linux@about/essy/indecisive/TD-Linux) |
| 21:00.06 | CIA-24 | BZFlag: 03blast007 * r15888 10/trunk/bzflag/src/bzfs/bzfs.cxx: Fix the return values of getAPIMaxWaitTime |
| 21:00.19 | JeffM | what was wrong with it? |
| 21:02.13 | CIA-24 | BZFlag: 03blast007 * r15889 10/trunk/bzflag/plugins/nagware/nagware.cpp: Fix the call of bz_setMaxWaitTime |
| 21:02.23 | blast007 | JeffM: it generated a warning |
| 21:02.35 | blast007 | and it only returned a value if it was > 100, and when it was, it return 0 |
| 21:02.39 | brlcad | hm, i've run several profiles.. wonder which one he's referring to |
| 21:02.47 | blast007 | so it either returned 0, or nothing ;) |
| 21:02.50 | brlcad | the mysql queries, for example, are one aspect that are pretty bad |
| 21:03.09 | brlcad | phpbb logins seem to be particularly expensive, for example |
| 21:03.24 | brlcad | (not sure what aspect of the 'list server' you're referring to) |
| 21:04.53 | brlcad | he was probably referring to the rate profiles I'd done, which was more a measure of connections per second .. you can't exactly profile web apps the same way as code to see "where" exactly they spend a lot of time (e.g. in the php) other than which pages and which operations |
| 21:05.18 | JeffM | blast007, ahh |
| 21:05.40 | JeffM | brlcad, the things done by bzfls.php mostly |
| 21:05.44 | brlcad | most informative was the connection/query rate which .. off the top of my head is anywhere as high as about 20 per second or less .. with average of around 1 or 2 queries per second |
| 21:06.05 | JeffM | how expensive is just a list of servers, compared to a client get token, compared to a check token, etc.. |
| 21:06.12 | brlcad | sometimes less of course, but average seems to be about 1/sec with spikes around 20/sec |
| 21:06.51 | JeffM | I'm curious if doing things like caching the list of servers will help, so we'd not have to query sql for that, only build it on an add or remove. |
| 21:06.55 | *** join/#bzflag LongDon (n=LongDon@host-091-097-077-012.ewe-ip-backbone.de) |
| 21:09.45 | blast007 | I would imagine that switching between two databases would be slower than having a single database |
| 21:09.55 | *** join/#bzflag orthox (n=orthox@70.89.111.225) |
| 21:11.18 | brlcad | JeffM: I've not really seen the server list query ever show up as being an expensive call |
| 21:11.28 | brlcad | the table is utterly tiny in comparison to the user table |
| 21:11.47 | *** part/#bzflag Tedius (n=Tedius@musictech.MUSC.CWRU.Edu) |
| 21:11.53 | brlcad | user logins and token validation are fairly expensive |
| 21:12.23 | brlcad | anything that hits the user tables really .. searching for or listing players on the website, for example |
| 21:13.04 | brlcad | it only takes a half dozen of those to clog it up so that everyone starts getting held up in the queue |
| 21:13.19 | JeffM | l |
| 21:13.20 | JeffM | k |
| 21:13.38 | JeffM | wonder if it makes in-eficent calls to it |
| 21:13.38 | brlcad | holding an in-memory cache of the person's password if they'd recently logged in would probably help a lot |
| 21:13.50 | JeffM | and group list as well I'd guess |
| 21:14.39 | brlcad | SELECT user_id FROM phpbb_users WHERE username='blahblah' AND user_password=MD5('blahblah') AND user_active=1; .. is pretty typical |
| 21:14.56 | *** join/#bzflag jude- (n=jude@wireless-outside-18.uawireless.Arizona.EDU) |
| 21:15.00 | brlcad | you could do the md5 in apache-land, but that doesn't really buy you anything |
| 21:15.07 | jude- | TGIF |
| 21:15.15 | brlcad | there's no table joins, or other silly mistakes |
| 21:15.18 | JeffM | k, mostly just curious |
| 21:15.27 | jude- | hello all |
| 21:15.39 | Constitution | hi jude- |
| 21:15.41 | brlcad | howdy jude- |
| 21:17.22 | brlcad | if you had a "password manager" of sorts (whether in php or c++ code) that hooked into the db, and all queries went through it, it could keep a MRU list of usernames (like the last 500 in a hash) |
| 21:17.39 | brlcad | i'd expect that to be better than always searching through a list of 15000 players |
| 21:19.51 | JeffM | yeah |
| 21:19.59 | JeffM | cus I'd guess ram usage isn't high on the machine |
| 21:21.02 | JeffM | the problem is for passwords, what do you do if the user changes the password in bzbb, then somone goes and tries to login using his hold password |
| 21:21.12 | JeffM | like he let it slip out |
| 21:21.20 | JeffM | says "crap" goes and changes it |
| 21:21.35 | JeffM | then some jerk loges in using his old password, we cached it, and the jerk gets in |
| 21:22.23 | JeffM | what'd be nice is if the management console for users (phpbb in this case ) could send the list the bzid to mark as dirty. |
| 21:22.29 | JeffM | independent of sql |
| 21:23.23 | Manu | brlcad: I guess phpbb_users has an index for the fields in the "where" part, right? |
| 21:28.30 | tannerld | ~blast007++ |
| 21:28.39 | tannerld | ~SpazzyMcGee++ |
| 21:30.14 | SpazzyMcGee | ~tannerld++ |
| 21:30.22 | SpazzyMcGee | you need it too. |
| 21:30.26 | tannerld | heh |
| 21:30.35 | tannerld | "technically" I should have more than that :P |
| 21:30.50 | SpazzyMcGee | "technically" beggars can't be choosers. |
| 21:30.56 | SpazzyMcGee | :P |
| 21:31.10 | tannerld | heh |
| 21:37.32 | *** join/#bzflag KTL (n=KTL@213.219.188.29.adsl.dyn.edpnet.net) |
| 21:38.32 | forrestv | lol |
| 21:38.35 | forrestv | what is wtf? |
| 21:38.44 | JeffM | google it |
| 21:38.48 | forrestv | ibot: what is wtf? |
| 21:38.49 | ibot | ACTION is wtf to understand recursion, you must first understand recursion |
| 21:39.34 | brlcad | JeffM: we're okay ram-wise .. it's got just a GB and where almost always below it even with 20 "spare" httpds running (which eat up 20-30MB each) |
| 21:40.07 | JeffM | so a couple megs to cache out stuff would not be a killer ;) |
| 21:40.14 | brlcad | the passwords are a problem, that's why I mentioned having all passwords go through some central password manager .. so even phpbb would update the cache |
| 21:40.32 | JeffM | ahh |
| 21:40.42 | brlcad | not saying it's easy or a good way, but it's a way ;) |
| 21:40.44 | JeffM | well if we could do that, then we'd just have phpbb use ldap :) |
| 21:41.17 | brlcad | with ldap and a live daemon, ldap could actually notify the daemon when the password changed iirc |
| 21:41.22 | Manu | brlcad: what's the main problem, cpu usage, query speed...? |
| 21:42.02 | brlcad | Manu: combination and depends what's going on -- mostly query speed .. which is of course tied to cpu speed |
| 21:42.26 | Manu | mysql is known for his cpu usage |
| 21:42.27 | brlcad | the cpu isn't new, but it's not ancient crap either .. what we're trying to do should be *way* under its performance specs |
| 21:42.36 | brlcad | yeah, mysql is the pig atm |
| 21:42.43 | brlcad | handling a ton of queries |
| 21:42.54 | brlcad | so when mysql gets too busy, everything gets backed up |
| 21:42.56 | Manu | and I guess you are using the right indexes for perfomance, right? |
| 21:43.10 | Manu | I haven's seen the database schema |
| 21:43.29 | Manu | 50 queries per second is nothing big |
| 21:43.37 | brlcad | so we either improve mysql performance, improve query performance, avoid the queries altogether with other optimizations, or get a faster processor :) |
| 21:43.53 | Manu | or change the database ;) |
| 21:43.58 | brlcad | i'm not doing anything .. we're not doing anything .. we use what phpbb does |
| 21:44.10 | Manu | ah, that's the problem :( |
| 21:44.20 | Manu | probably is not designed for the use we do |
| 21:44.25 | brlcad | probably |
| 21:52.50 | *** join/#bzflag menotume (n=menotume@pdpc/supporter/active/menotume) |
| 21:52.51 | *** mode/#bzflag [+v menotume] by ChanServ |
| 22:00.06 | CIA-24 | opencombat: 03jeffm2501 * 10superList/ (12 files in 3 dirs): silly stand alone web server. |
| 22:02.02 | JeffM | do we still need the pre-mesh tag? |
| 22:02.16 | JeffM | or even the 1.11 ones? |
| 22:12.31 | Erroneous | doesn't hurt anything |
| 22:12.41 | Erroneous | 'cept the silly folks who try to check out the entire repo |
| 22:12.56 | JeffM | 380 megs and counting ;) |
| 22:13.20 | Erroneous | why check it all out? |
| 22:13.29 | JeffM | I missclickec |
| 22:13.33 | Erroneous | heh |
| 22:13.43 | JeffM | I wanted all of trunk |
| 22:13.59 | *** join/#bzflag Soraya_ (i=439e755b@gateway/web/cgi-irc/zeebrothers.net/x-13da593409884979) |
| 22:14.00 | *** part/#bzflag Soraya_ (i=439e755b@gateway/web/cgi-irc/zeebrothers.net/x-13da593409884979) |
| 22:15.52 | *** join/#bzflag Legoguy_ (n=Legoguy@70.142.201.242) |
| 22:27.13 | brlcad | another pig of an sql call that is fairly common: SELECT callsign,highscore,highstrengthindex FROM `playerinfo` ORDER BY `highscore` DESC LIMIT 20; |
| 22:27.32 | Erroneous | ah, stats |
| 22:28.01 | brlcad | often take 10-20 seconds per query, searching about 350k rows |
| 22:28.40 | Manu | wow.. incredibly slow |
| 22:28.40 | KTL | ah that db, good that someone looks at it |
| 22:29.00 | brlcad | another one that I'm not sure there's a really good alternative other than hashing some indexes and using stored queries |
| 22:29.37 | brlcad | so that query, for example, would always be solved |
| 22:29.46 | Manu | brlcad: so probably queries aren't optimized |
| 22:31.27 | brlcad | the query itself is simple .. there's not much to optimize when you've got a table that big and you want the top N sorted by some particular row (which is user-selectible) .. other than telling mysql to always have the answer to that query ready in a stored result |
| 22:31.49 | Manu | brlcad: you can optimize it adding indexes |
| 22:33.28 | brlcad | that would be part of the "telling mysql to always have the answer" ;) |
| 22:33.38 | Manu | hehe :) |
| 22:34.12 | KTL | and rebuild that index every so many days |
| 22:34.44 | Manu | KTL: rebuild? it may be updated everytime a record is added/modified/deleted |
| 22:38.05 | KTL | yes ok some of my old code had to do it because the whole content could have been changed everytime the program ran. |
| 22:38.56 | Manu | yes, for that case is better disable index and then rebuild in one step |
| 22:39.53 | KTL | in fact i kept a database of files. And I did not know of an easy way to update the database the moment the filesystem changed |
| 22:40.52 | KTL | he that database, i believe it would stimulate competition if there where weekly statistics or something |
| 22:41.05 | KTL | instead of only those "all time totally screwed up stats" |
| 22:43.31 | brlcad | are you willing to make the mods? |
| 22:43.47 | KTL | aw there i go again |
| 22:44.09 | KTL | it's php right? |
| 22:45.07 | KTL | my attention span is too short for long lasting help i'm afraid |
| 22:53.05 | Manu | no harm with indexes... I guess ;) |
| 22:53.20 | Manu | but who knows... |
| 22:53.56 | KTL | Indexes speed up retrieval but slow down inserts, deletes and updates of values in indexed columns. |
| 22:54.46 | brlcad | ah, sweet |
| 22:55.02 | brlcad | down to < 1 sec now on all the stats site queries |
| 22:55.22 | Manu | you mean total time < 1 seg? |
| 22:55.24 | Manu | sec* |
| 22:55.37 | KTL | nice nice :) |
| 22:56.26 | brlcad | yep |
| 22:56.28 | brlcad | 0.02s |
| 22:56.34 | brlcad | from 10+sec |
| 22:56.36 | blast007 | heh |
| 22:56.38 | Constitution | wow |
| 22:57.07 | Manu | brlcad: indexes did the magic... and your fingers of course ;) |
| 22:57.09 | donny_baker | KTL: True, but the records are retreived more often than updated so the net effect is a positive effect on performance |
| 22:57.15 | brlcad | only put an index on the four tables that the player stats page reports on |
| 22:57.23 | brlcad | dunno what the insert impact will be now though |
| 22:57.28 | brlcad | s/tables/columns/ |
| 22:57.32 | Manu | too low brlcad :) |
| 22:57.40 | brlcad | hopefully |
| 22:57.44 | Manu | except if u insert 1000 tuples at the same time ;) |
| 22:57.55 | Manu | 10000+ |
| 22:58.08 | brlcad | now, http://stats.bzflag.org/playerstats.php reports instantly .. woot |
| 22:58.36 | KTL | impossibly high scores :( |
| 22:58.40 | blast007 | wow, the player search is a *lot* faster too |
| 22:58.40 | Manu | remember I told you :p |
| 22:58.55 | tannerld | ~brlcad++ |
| 23:00.29 | donny_baker | yeah, a few well placed indexes do wonders on db performance, insterts take a hit, but I've never regretted putting a *needed* index on a table |
| 23:01.32 | *** part/#bzflag LongDon (n=LongDon@host-091-097-077-012.ewe-ip-backbone.de) |
| 23:02.24 | KTL | brlcad, what is the db scheme used? how does it look like? |
| 23:02.54 | KTL | to have an idea about what more could be possibly done with the data |
| 23:03.25 | KTL | gn |
| 23:05.02 | brlcad | I think I took care of invalid username queries that were taking 5-10 seconds |
| 23:05.13 | brlcad | global logins, that is |
| 23:05.26 | brlcad | <PROTECTED> |
| 23:07.25 | brlcad | there are indices on the user table .. but none on any of the columns we were querying during a player lookup |
| 23:07.52 | brlcad | hm.. hopefully that will speed up global logins significantly |
| 23:07.55 | brlcad | ~manu++ |
| 23:08.51 | blast007 | brlcad: definately does |
| 23:08.59 | blast007 | several times faster |
| 23:09.13 | brlcad | woot |
| 23:10.18 | blast007 | 1.25 seconds to get the list in 2.0.9, along with a global token |
| 23:10.50 | blast007 | might be somewhat impacted because I run it though a local script as sort of a proxy |
| 23:11.21 | brlcad | and now? |
| 23:11.25 | brlcad | or before |
| 23:11.47 | blast007 | before it was probably closer to 5 seconds |
| 23:13.58 | brlcad | could be better, but not too shabby |
| 23:14.22 | *** join/#bzflag Soraya_ (i=439e755b@gateway/web/cgi-irc/zeebrothers.net/x-886df779d68ebd86) |
| 23:14.23 | brlcad | maybe that'll mask that annoying bug where the login takes too long and you end up not getting a token |
| 23:14.23 | *** part/#bzflag Soraya_ (i=439e755b@gateway/web/cgi-irc/zeebrothers.net/x-886df779d68ebd86) |
| 23:16.02 | blast007 | more like 1.10 seconds when I access the list + global auth in a web browser |
| 23:18.10 | brlcad | wow, mysqld dropped below 1% use |
| 23:19.15 | brlcad | it's usually 3-20%, lingering around 5% most of the time |
| 23:19.37 | brlcad | ~bzflist |
| 23:19.52 | brlcad | so that means we can take 500 players now *ahem* |
| 23:20.10 | *** join/#bzflag rob1n (n=emp@unaffiliated/rob1n) |
| 23:31.07 | donny_baker | forums seem more responsive too :) |
| 23:31.27 | brlcad | maybe the index on username |
| 23:32.19 | donny_baker | or just not having to wait for the engine on the other queries |
| 23:36.05 | brlcad | yeah, probably that more than anything |
| 23:38.43 | *** join/#bzflag bryjen (n=bryjen@2002:4cb1:bee3:0:0:0:0:1) |
| 23:38.43 | *** mode/#bzflag [+v bryjen] by ChanServ |