| 00:11.15 | blast007 | Erroneous: should there be an issue if the string passed to regexec is blank? in the instance it is crashing, I am not passing a command line argument to the plugin |
| 00:17.07 | blast007 | and it does not crash in 2.0.x, just 2.3... |
| 00:18.55 | Erroneous | blast007: blank string should be fine. pointer can't be null, string must be null-terminated, but zero-length should be fine. |
| 00:19.07 | blast007 | hrm |
| 00:19.24 | blast007 | (gdb) print commandLine |
| 00:19.24 | blast007 | $1 = 0x7ffff7342158 "" |
| 00:19.24 | blast007 | (gdb) print *commandLine |
| 00:19.24 | blast007 | $2 = 0 '\000' |
| 00:19.46 | blast007 | yet it crashes when calling regfree -> free |
| 00:20.00 | blast007 | and the result of regcomp is 0 |
| 00:20.40 | blast007 | the result of regexec is 1 (no match) |
| 00:21.20 | Erroneous | if regcomp succeeds, regfree should not crash at all |
| 00:21.26 | Erroneous | even if you never call regexec |
| 00:55.13 | *** join/#bzflag TimRiker (~TimRiker@bzflag/projectlead/TimRiker) |
| 00:55.13 | *** mode/#bzflag [+o TimRiker] by ChanServ |
| 01:04.15 | *** join/#bzflag Anxuiz (~Anxuiz@unaffiliated/anxuiz) |
| 01:22.22 | CIA-5 | BZFlag: 03McSpider 07http://my.bzflag.org * r7790 10/w/Bz_fireWorldWep: Added missing parameter, Changed page style to match newer pages |
| 01:22.40 | CIA-5 | BZFlag: 03McSpider 07http://my.bzflag.org * r7791 10/w/Functions%28API%29: Added missing parameter |
| 02:19.08 | CIA-5 | BZFlag: 03Thumper 07http://my.bzflag.org * r7792 10/w/BZFS_API_2.4_Upgrade: typo fix |
| 02:30.06 | CIA-5 | BZFlag: 03Thumper 07http://my.bzflag.org * r7793 10/w/BZFS_API_2.4_Upgrade: Fix names for 2.0.x register event and remove event methods |
| 02:38.17 | *** join/#bzflag Anxuiz (~Anxuiz@50-46-121-138.evrt.wa.frontiernet.net) |
| 02:38.17 | *** join/#bzflag Anxuiz (~Anxuiz@unaffiliated/anxuiz) |
| 05:11.10 | *** join/#bzflag spldart (~spldart2@bzflag/contributor/spldart) |
| 05:11.10 | *** mode/#bzflag [+v spldart] by ChanServ |
| 05:38.28 | BulletCatcher | blast007: FWIW, the TimeLimit plugin works normally for me at BRL.ARPA.NET:15157. |
| 05:38.35 | BulletCatcher | -loadplugin TimeLimit,15,20,30 |
| 05:38.43 | BulletCatcher | is how I invoke it |
| 05:59.22 | BulletCatcher | Yay, Constitution's RC1 .app works fine on my 10.4 MacBook. |
| 05:59.28 | BulletCatcher | ~Constitution++ |
| 06:25.06 | *** join/#bzflag Marzipan- (~Marzipan@p5B222246.dip.t-dialin.net) |
| 06:26.37 | *** join/#bzflag Marzipan (~Marzipan@bzflag/player/Marzipan) |
| 06:47.02 | Constitution | ~celebrate |
| 06:47.02 | ibot | ACTION detonates a small nuclear device to celebrate the end of the world |
| 06:52.31 | Constitution | but hell has frozen over so we can go there now |
| 07:10.55 | ruskie | it has? |
| 07:26.00 | Constitution | yes, since duke nukem and bzflag have actually released (well, almost) |
| 07:36.13 | ruskie | rofl |
| 07:55.34 | *** join/#bzflag AAA_awright_ (~a3@ip24-251-157-63.ph.ph.cox.net) |
| 08:17.15 | *** join/#bzflag Constitution (~const@bzflag/developer/Constitution) |
| 08:20.26 | *** join/#bzflag Pimpinella (~frank@gondolin.pimpi.org) |
| 08:21.54 | *** join/#bzflag Pimpi (~frank@gondolin.pimpi.org) |
| 08:37.58 | brad | got bz installed in a netbook......but ive noticed quite a lot of the time one tank on radar is yellow not purple/red. what does that symbolize? |
| 08:40.26 | brad | http://moooo.org/img/0/e93af44.png |
| 08:53.07 | *** join/#bzflag Guest406 (2e15d71d@gateway/web/freenode/ip.46.21.215.29) |
| 08:58.02 | Constitution | ~lart brad |
| 08:58.02 | ibot | pours hot grits down the front of brad's pants |
| 08:58.18 | brad | ~bzfrag Constitution |
| 08:58.18 | ibot | ACTION destroys Constitution with a guided missile |
| 08:58.24 | brad | did you win that game? :p |
| 08:58.32 | Constitution | no |
| 08:58.35 | brad | :( |
| 08:58.42 | Constitution | ~spell squerril |
| 08:58.49 | Guest406 | breado woke up :P |
| 08:58.53 | Constitution | yeah, that guy... |
| 08:58.59 | brad | ziggy is that you |
| 08:59.06 | Guest406 | who else:P |
| 08:59.10 | brad | :) |
| 08:59.12 | brad | hey |
| 08:59.13 | Guest406 | ;) |
| 08:59.17 | Guest406 | nheya |
| 08:59.26 | Guest406 | raining morning;) |
| 08:59.33 | brad | so cons any idea about the weird yellow tank? :p |
| 08:59.56 | brad | ~weather egss |
| 09:00.53 | *** join/#bzflag zgrywus (2e15d71d@gateway/web/freenode/ip.46.21.215.29) |
| 09:07.56 | blast007 | BulletCatcher: load TimeLimit without anything after it. Just -loadplugin TimeLimit |
| 09:08.05 | blast007 | it does not require an option to be passed |
| 09:13.15 | blast007 | I mean.. it could be fixed by just not running the regex stuff if there is nothing passed in |
| 09:13.49 | blast007 | but it works fine in 2.0.x, so *something* is different with 2.3 for how that works |
| 09:43.23 | Constitution | brad: not really, unless your network is so messed up that it's corrupting packets |
| 09:43.48 | Constitution | or, http://www.apple.com/why-mac |
| 09:44.40 | blast007 | yeah, get a Mac - then the occasional yellow tank on radar will be the least of your concern! |
| 09:45.11 | blast007 | welcome to the walled garden |
| 09:45.57 | Constitution | don't be jealous |
| 09:46.06 | blast007 | don't worry - I'm not ;) |
| 09:46.15 | Constitution | mhm :) |
| 09:47.42 | blast007 | have you bought your $50 thunderbolt cable yet? |
| 09:50.19 | Constitution | you can complain about that once HDMI cables come down in price :) |
| 09:50.59 | Constitution | since those are just actual cables (afaik), and thunderbolt cables have chips and circuitry and everything |
| 09:51.22 | blast007 | $11 for an HDMI cable? |
| 09:53.39 | Constitution | um... I wish I had your electronics store |
| 09:53.47 | blast007 | the internet |
| 09:54.00 | blast007 | http://www.cablestogo.com/product.asp?cat_id=3603&sku=40303 |
| 09:54.26 | Constitution | hm, I've been ripped off! |
| 09:55.03 | blast007 | I just think that thunderbolt will die off, and then you'll need to buy a new Mac to get USB 3.0 ;) |
| 09:58.15 | delusional | thunderbolt is twice as fast as usb 3 |
| 10:01.02 | *** join/#bzflag nielsle (~nielsle@4135136-cl69.boa.fiberby.dk) |
| 10:06.02 | blast007 | delusional: doesn't mean anything ;) firewire was generally faster than USB, and look what happened to that |
| 10:07.28 | delusional | video professionals need something that provides some wattage |
| 10:07.45 | delusional | it's aimed at them |
| 10:09.30 | ruskie | well intel is known for pushing bad tech successfully(they've done it so far with x86 instead of a better arch) |
| 10:11.11 | delusional | it's not bad tech.... it's just might never become ubiquitous |
| 10:29.30 | blast007 | someone said that some Sony VAIO systems will have it as well, but it will be a proprietary version... |
| 10:29.45 | blast007 | so you'll get thunderbolt-like ports that aren't standard :) |
| 10:30.11 | blast007 | fragmenting the market and there's not even any products released quite yet.. |
| 10:37.20 | ruskie | so basically standard practise |
| 11:16.31 | delusional | still, I will have something that you can't get. |
| 11:17.54 | ruskie | ? |
| 11:18.17 | delusional | not you. |
| 11:18.20 | *** join/#bzflag trepan (~trepan@unaffiliated/trepan) |
| 11:18.20 | *** mode/#bzflag [+v trepan] by ChanServ |
| 11:57.47 | *** join/#bzflag Zac11 (ca59b013@gateway/web/freenode/ip.202.89.176.19) |
| 12:06.25 | *** join/#bzflag mdskpr_ (~mdskpr@108.25.145.104) |
| 12:24.38 | CIA-5 | BZFlag: 03Blast 07http://my.bzflag.org * r7794 10/w/Getting_Help: |
| 12:25.25 | *** join/#bzflag saxotest (~chatzilla@216.252.79.92) |
| 13:32.40 | blast007 | BulletCatcher: hmm, and mine works with -loadplugin TimeLimit,15-30, but not with TimeLimit,15,20,30 nor TimeLimit |
| 13:32.46 | blast007 | the latter two crash |
| 13:33.17 | blast007 | are we only supposed to run regfree if there was a match...? |
| 13:33.50 | blast007 | #0 0x00007ffff6ab8e2d in free () from /lib/libc.so.6 |
| 13:33.50 | blast007 | #1 0x00007ffff6af95da in regfree () from /lib/libc.so.6 |
| 13:33.51 | blast007 | #2 0x00007ffff36e8542 in isValidCmdLine (regex=0x7ffff36e9298 "^[0-9]+-[0-9]+$", commandLine=0xa04288 "15,30") at TimeLimit.cpp:66 |
| 13:34.05 | blast007 | (line number is a bit difference since I had added a check on regcomp) |
| 13:34.31 | blast007 | 66 is the call to regfree |
| 13:34.40 | blast007 | is anyone else getting this crash? |
| 13:35.16 | blast007 | both of these crash for me: bzfs -loadplugin TimeLimit or bzfs -loadplugin TimeLimit,15,20,30 |
| 13:36.09 | blast007 | in trunk |
| 13:36.24 | blast007 | it works just fine in 2.0.x |
| 13:53.12 | trepan | blast007: Upon successful completion, the regcomp() function shall return 0. Otherwise, it shall return an integer value indicating an error as described in <regex.h>, and the content of preg is undefined |
| 13:53.22 | trepan | that last 7 words are the interesting ones |
| 13:53.44 | blast007 | yeah, regcomp is the one I am making sure returns a 0, and if not, I have it return out of the function |
| 13:53.46 | trepan | (that's from 'man 3posix regexec') |
| 13:54.20 | trepan | blast007: ah, the code currently ignores it |
| 13:54.38 | blast007 | my local copy doesn't |
| 13:54.54 | blast007 | but I was already checking the return value in the debugger and result was being set to 0 |
| 13:55.50 | blast007 | so it IS compiling correctly, and regexec DOES run (I get a return of 1 meaning there is no match), but regfree crashes |
| 13:56.02 | trepan | <PROTECTED> |
| 13:56.15 | blast007 | and this same code in 2.0.x does not crash |
| 13:56.32 | trepan | built-in vs. system regex? |
| 13:56.41 | blast007 | not sure |
| 13:58.11 | blast007 | how would I check that? |
| 13:58.30 | trepan | config.log? |
| 13:59.02 | blast007 | 2.4 has it, but 2.0.x does not show that information |
| 13:59.10 | blast007 | for 2.4, it is using the system regex |
| 13:59.14 | trepan | 'nm -D', maybe? |
| 14:01.08 | blast007 | hmm, for 2.4 the reg* symbols are undefined, and for 2.0.x they are "in the text (code) section." |
| 14:01.36 | trepan | heh, that is a little strange, my 2.0.x exec is using the built-in regex |
| 14:01.42 | trepan | don't know what triggered it |
| 14:02.58 | trepan | hm, it's the default |
| 14:03.23 | blast007 | I imagine if I forced the regex build it would work, then :) |
| 14:03.33 | trepan | ya, probably |
| 14:03.44 | trepan | 2.4 inverted the BUILD_REGEX default |
| 14:04.22 | trepan | but still, if the regex calls are being used correctly, there shouldn't be a crash (regardless of which lib is being used) |
| 14:07.08 | blast007 | unless we're maybe including some other file from BZFlag's tree that conflicts with the system regex |
| 14:13.16 | trepan | ya, bzregex.h never #includes regex.h, and TimeLimit.cxx #includes it directly (whereas common.h tests against BUILD_REGEX, as does motd.cxx; all very messy) |
| 14:14.32 | trepan | bzregex.h should probably do the BUILD_REGEX check, would be simpler |
| 14:29.16 | BulletCatcher | I agree. For now, TimeLimit.cpp should contain a copy of the #ifdef section from the bottom of common.h . |
| 14:30.32 | Pimpinella | anyone familiar with building RPM packages? |
| 14:30.59 | blast007 | Pimpinella: first, find a goat and a knife. |
| 14:31.16 | Pimpinella | ? |
| 14:31.32 | cygal | Pimpinella: it's a bit arcane :) |
| 14:31.37 | BulletCatcher | And after 2.4.0 I'll change motd.cxx to include common.h rather than having its own variation of that same #ifdef stuff. |
| 14:31.40 | blast007 | Pimpinella: for the sacrifice |
| 14:31.45 | Pimpinella | ah |
| 14:31.46 | mdskpr_ | and coat a hat in some blood to make a red hat |
| 14:32.21 | blast007 | BulletCatcher: k, let me try adding that ifdef |
| 14:32.36 | Pimpinella | is happy to use a .deb based distro |
| 14:33.35 | blast007 | on debian you need a pig and a mallet. On Windows, you need a mouse and a keyboard.. well, and technically you don't even really NEED the mouse. |
| 14:33.41 | Pimpinella | that probably means we will not be able to make ready to use packages for susi and fedora? |
| 14:34.07 | blast007 | BulletCatcher might be able to - or at least, I know he *uses* SUSE ;) |
| 14:34.09 | BulletCatcher | Pimpinella: I have done RPMs, but the .spec file language is a challenge. |
| 14:34.25 | BulletCatcher | I use Fedora. |
| 14:34.30 | blast007 | close enough |
| 14:34.33 | BulletCatcher | :-p |
| 14:34.47 | blast007 | yum install cookies yum yum |
| 14:35.09 | Pimpinella | i've made .debs, whch is a pain but at least well documented |
| 14:35.39 | blast007 | we could/should talk to the package maintainers of various distros |
| 14:37.25 | BulletCatcher | I'll be in touch with the Fedora package maintainer when the time comes. |
| 14:38.45 | BulletCatcher | I hope to figure out how to create a 2.4 RPM that can be installed at the same time with the 2.0.16 RPM. |
| 14:38.47 | blast007 | do most distros generally not release a new version of software in their existing releases? |
| 14:39.21 | Pimpinella | i don't think so |
| 14:39.29 | BulletCatcher | I'm sure it varies. |
| 14:39.33 | blast007 | yeah |
| 14:40.44 | *** join/#bzflag Upsetter (~er@i59F6CB75.versanet.de) |
| 14:42.11 | BulletCatcher | In our case, the fundamental incompatibility of 2.4 with 2.0 should encourage them to provide the new version. |
| 14:42.15 | BulletCatcher | What we can't control at all is their schedule. |
| 14:43.18 | brad | ?works here |
| 14:43.22 | brad | oops |
| 14:43.31 | brad | i was scrolled way up |
| 14:43.32 | brad | ignore me |
| 14:43.45 | Pimpinella | ignores brad |
| 14:43.48 | brad | :D |
| 14:44.18 | blast007 | did brad say something? |
| 14:44.33 | blast007 | I must have him on ignore |
| 14:44.57 | brad | ~ignore me |
| 14:44.57 | ibot | ACTION sticks her fingers in her ears. "La, la, la! I can't hear you, brad!" |
| 14:45.02 | blast007 | ;) |
| 14:48.10 | *** join/#bzflag bier|tp (~bier@p57A66A26.dip.t-dialin.net) |
| 14:49.06 | *** join/#bzflag bier (~bier@p57A66A26.dip.t-dialin.net) |
| 14:51.44 | *** join/#bzflag TimRiker (~TimRiker@bzflag/projectlead/TimRiker) |
| 14:51.44 | *** mode/#bzflag [+o TimRiker] by ChanServ |
| 15:05.32 | *** join/#bzflag spldart (~spldart2@c-98-201-137-215.hsd1.tx.comcast.net) |
| 15:05.32 | *** join/#bzflag spldart (~spldart2@bzflag/contributor/spldart) |
| 15:05.32 | *** mode/#bzflag [+v spldart] by ChanServ |
| 15:41.34 | blast007 | BulletCatcher: seems like I'd also need to include config.h in TimeLimit? |
| 15:42.57 | blast007 | nvm, that does seem to do the trick |
| 15:43.06 | BulletCatcher | Good point, but up until now, at least, the plugins have been independent of that. |
| 15:43.17 | BulletCatcher | thinks about it |
| 15:43.45 | blast007 | I mean, it DOES work for now |
| 15:43.52 | BulletCatcher | Understood. |
| 15:44.01 | *** join/#bzflag JeffM (~JeffM@adsl-75-16-90-217.dsl.irvnca.sbcglobal.net) |
| 15:44.01 | *** join/#bzflag JeffM (~JeffM@unaffiliated/jeffm2501) |
| 15:44.01 | *** mode/#bzflag [+v JeffM] by ChanServ |
| 15:44.08 | blast007 | at least on Linux - haven't tried it on Windows with this change ;) |
| 15:44.25 | BulletCatcher | Windows always uses the embedded regex, so shouldn't be a problem. |
| 15:44.37 | blast007 | k |
| 15:44.56 | blast007 | the other option would be to just put the regex check back the way it was and default to using built-in |
| 15:45.25 | blast007 | but it would still break if someone forced it.. |
| 15:46.02 | BulletCatcher | I am leaning toward just including config.h, but I want a few more minutes to contemplate. |
| 15:46.07 | blast007 | k |
| 15:47.17 | blast007 | maybe for 2.4.2 we should expose some regex stuff via the BZFS API and just use that for plugins |
| 15:47.55 | JeffM | dont use config.h in plugins |
| 15:48.16 | JeffM | not everything in bzfs is exported on windows |
| 15:48.27 | JeffM | so if you need something, export it |
| 15:49.24 | blast007 | JeffM: the TimeKeeper plugin uses the reg* functions for regex matching right now, and it's breaking because it's using the bz regex while other stuff is using the system regex (or something like that) |
| 15:49.40 | JeffM | yeah that's a crappy plugin |
| 15:49.43 | JeffM | we should expose ours |
| 15:49.45 | blast007 | sorry, TimeLimit |
| 15:50.02 | blast007 | yeah |
| 15:50.05 | JeffM | export ours right now, what functions does it need? it should not take more then 10 min |
| 15:50.29 | blast007 | it calls regcomp, regexec, and regfree |
| 15:50.43 | JeffM | you wanna do it? or should I? |
| 15:51.09 | blast007 | how would you want it done? a single "does it match" function? |
| 15:51.29 | JeffM | if that's all it does then yeah |
| 15:51.34 | JeffM | higher level is better |
| 15:51.46 | JeffM | cus if we change to something else on the backend later the API dosn't change |
| 15:51.52 | blast007 | so far it just cares if it matches |
| 15:52.14 | BulletCatcher | Rather than make such a big change now, I am willing to force the use of the embedded regex for 2.4 and do a better fix later. |
| 15:52.25 | JeffM | that's not a big change BulletCatcher it's just one API function |
| 15:52.27 | JeffM | called by one plugin |
| 15:52.29 | BulletCatcher | That will make it more like 2.0. |
| 15:52.31 | JeffM | the risk is minimal |
| 15:52.47 | JeffM | I'd rather do that then change the build system for bzflag |
| 15:52.56 | JeffM | we know the bz build sysem is working now |
| 15:53.00 | BulletCatcher | It'll be an easy change. :-) |
| 15:53.10 | JeffM | but it affects more people |
| 15:53.17 | JeffM | this is also an easy change and affects less people |
| 15:53.21 | BulletCatcher | k |
| 15:53.33 | JeffM | I'm worried about invlidating the testing we have done |
| 15:53.34 | blast007 | and if someone enables the system regex, it'll still break ;) |
| 15:53.48 | blast007 | and some package maintainers might do that to cut down on the size |
| 15:54.04 | JeffM | or there might be other reasosn they need to use system libs |
| 15:54.10 | BulletCatcher | The problem we have right now is the use of system regex library with the embedded .h files. |
| 15:54.24 | JeffM | well that isn't good |
| 15:54.40 | JeffM | it all depends on what you mean by "force" |
| 15:54.42 | BulletCatcher | It is only a problem in the TimeLimit plugin. |
| 15:54.54 | JeffM | do you mean remove the option? or just default it to use internal? |
| 15:55.05 | BulletCatcher | Just change the default. |
| 15:55.13 | JeffM | then that dosn't solve the problem |
| 15:55.18 | JeffM | just gets around it for some people |
| 15:55.24 | JeffM | not a fan of that as a "fix" |
| 15:55.40 | BulletCatcher | In 2.0 it always builds the embedded regex. In 2.3 it will use the system regex by default. |
| 15:56.05 | JeffM | you say "always" for 2.0 there is no way to force the build system to use system regex? |
| 15:56.15 | BulletCatcher | looks |
| 15:56.44 | blast007 | 2.0.x has --with-regex Use bzflag regex lib |
| 15:56.50 | JeffM | because not having the option is always, having the option but defaulting to ours is not always, it's just mostlys |
| 15:57.12 | JeffM | and it's not good to have a build system option that will break something we know about |
| 15:57.39 | blast007 | and you can probably do --with-regex=no on 2.0.x |
| 15:57.47 | blast007 | if I'm reading configure.ac right ;) |
| 15:58.45 | JeffM | no mater what we need to expose an API function to do a regex |
| 15:59.06 | JeffM | our standard plugins should not be using third party libs |
| 16:00.34 | BulletCatcher | On *nix systems these days, regex is as standard as the math library. |
| 16:00.43 | JeffM | when using system regex in 2.3 does bzfs use our header or the system header? |
| 16:00.53 | JeffM | BulletCatcher, linux isn't the only OS ;) |
| 16:01.20 | JeffM | if you wanted to desuport servers on windows, I probably would not fight you very hard |
| 16:01.26 | JeffM | but then I'd do less development too |
| 16:01.46 | BulletCatcher | We don't need to go that far at this time. |
| 16:02.08 | JeffM | it is not a widely used OS for hosting |
| 16:02.16 | JeffM | probably the only use it gets is when I test localy |
| 16:02.22 | JeffM | and that ain't very often |
| 16:02.26 | BulletCatcher | 2.3 uses the system header with the system regex library. |
| 16:02.38 | BulletCatcher | (see the bottom of include/common.h) |
| 16:02.50 | JeffM | ok good, so all we have to do is make the plug-in not use regex and we are good |
| 16:02.54 | JeffM | that should be a quick fix |
| 16:03.00 | JeffM | blast007, you want to do it or should I? |
| 16:03.08 | BulletCatcher | Yes, that will be an adequate fix. |
| 16:03.23 | JeffM | not just adequate, correct ;) |
| 16:03.48 | JeffM | or at least consistent with all the other stuff we did in the API |
| 16:04.02 | JeffM | we had similar issues with stuff like toupper and tolower and tokenize |
| 16:04.27 | BulletCatcher | My only concern it that we'll define an inadequate API for the future, but right now the only need we have is for TimeLimit.h. |
| 16:04.34 | BulletCatcher | So go for it. |
| 16:04.42 | JeffM | yeah we can ALWAYS add more API functions |
| 16:05.36 | BulletCatcher | An appropriately renamed version of isValidCmdLine() from TimeLimit.cpp will be fine. |
| 16:06.23 | JeffM | what would be the name, "QueryInString" or something? |
| 16:06.24 | blast007 | by "make it not use regex" you mean to move the regex calls to the API, correct? |
| 16:06.35 | JeffM | blast007, yes |
| 16:06.43 | JeffM | so that bzfs does the regex not the plugin |
| 16:07.07 | JeffM | bz_regularExpression ? |
| 16:07.25 | BulletCatcher | bz_regexMatch ? |
| 16:07.45 | JeffM | everyone in the world in all languages know what regex means? |
| 16:08.24 | blast007 | JeffM: it's a pretty regular expression (oh the pun, it burns) |
| 16:08.29 | JeffM | heh |
| 16:08.29 | BulletCatcher | I don't expect all of them to be able to understand "regular expression" either. |
| 16:08.36 | JeffM | k |
| 16:08.59 | BulletCatcher | But it looks like other API names are verbose. |
| 16:09.10 | blast007 | and should it just be a boolean return, or should I make it return the number of matches? (not sure what kind of data we even get back.. I'd have to check) |
| 16:09.14 | JeffM | yeah I am extra verbose |
| 16:09.26 | JeffM | the number sounds better |
| 16:09.31 | blast007 | bz_doesThisStringMatchThisRegularExpression |
| 16:09.32 | JeffM | and it can be used like a boolean |
| 16:09.58 | blast007 | okay, I'll see what I can do :) |
| 16:09.59 | JeffM | bz_expressionMatch |
| 16:10.05 | JeffM | oh I'm almost done with it :) |
| 16:10.13 | blast007 | oh :P |
| 16:10.53 | BulletCatcher | If it returns a number, then bz_regularExpressionMatches (plural) will help indicate that. |
| 16:11.36 | DTRemenak | what about match retrieval? consecutive queries with the same expression? replace? |
| 16:11.36 | BulletCatcher | That can also be interpreted as a verb, which isn't necessarily bad. |
| 16:11.50 | DTRemenak | why are we exporting library functions, badly? |
| 16:12.02 | DTRemenak | instead of just linking to the library? |
| 16:12.12 | JeffM | because the plugin dosn't link to the right library |
| 16:12.14 | blast007 | DTRemenak: because currently the plugin includes the built-in regex instead of the system regex |
| 16:12.19 | JeffM | you want to fix the plugin go for it |
| 16:12.59 | JeffM | overhual the build system to allow plugin make files to add references to any of our src/other libs so they can be pulled form the right place. |
| 16:13.06 | blast007 | could just make the plugin not need regex matching |
| 16:13.19 | JeffM | that would be the best way to support third party libs in plugins |
| 16:14.04 | BulletCatcher | I think I can rewrite the plugin not to use regex at all. |
| 16:14.08 | JeffM | do that then |
| 16:14.12 | JeffM | that's even less change |
| 16:14.30 | JeffM | then we can enhance the build system to let people use whatever regex is right later |
| 16:15.17 | DTRemenak | plugins are built from the same configure, right? should just be a matter of applying some BUILD_REGEX logic |
| 16:15.30 | JeffM | yeah I would not expect it to be that hard |
| 16:15.33 | blast007 | DTRemenak: then we have to include config.h in the plugin |
| 16:16.01 | blast007 | and that *did* work |
| 16:16.02 | JeffM | conditionaly |
| 16:16.03 | DTRemenak | blast007: since configure is for the entire project, that doesn't seem unreasonable |
| 16:16.17 | JeffM | just #def it for not windows an you'll be fine |
| 16:16.26 | DTRemenak | why not for windows? we've got one too |
| 16:16.27 | JeffM | we don't ship config |
| 16:16.35 | JeffM | you don't need bzfs source to build plugins |
| 16:16.55 | DTRemenak | msvc/build/config.h |
| 16:17.04 | blast007 | we should include that with the instaler |
| 16:17.05 | JeffM | we don't ship that in a bin install |
| 16:17.15 | JeffM | the bin install includes EVERYTING you need to build a plugin |
| 16:17.17 | DTRemenak | you aren't building timelimit in a bin install, it's already built |
| 16:17.28 | JeffM | no but there could be a pluign based on it that they have source for |
| 16:17.39 | blast007 | DTRemenak: what about 3rd party plugins that use it? or what if someone wants to tweak TimeLimit? |
| 16:17.40 | JeffM | is trying to be consistent |
| 16:17.42 | JeffM | we can ship it |
| 16:17.49 | JeffM | I think it's stupid cus it's not used |
| 16:17.52 | JeffM | but we can |
| 16:18.08 | blast007 | if we ship it, we'll have to monkey with the vcproj settings so it can find it ... |
| 16:18.22 | JeffM | I already did that |
| 16:18.28 | JeffM | environment vars FTW |
| 16:18.29 | DTRemenak | then they are responsible for including appropriate libraries. just like if I make a plugin that uses $RANDOM_LIBRARY, if someone wants to derive from that plugin, they also need $RANDOM_LIBRARY |
| 16:19.06 | JeffM | yes but that random libary is self contained, rooting around and finding the config in an MSVC folder in a bzflag source dir is not easy |
| 16:19.15 | JeffM | I can make it work |
| 16:19.18 | JeffM | we can ship it |
| 16:19.29 | DTRemenak | you don't need config.h if you have an external regex lib |
| 16:19.35 | DTRemenak | BUILD_REGEX will be false and everything will just work |
| 16:19.58 | JeffM | but to work with an internal regex you need it, so it has to beincluded in the source file for the plugin |
| 16:20.19 | JeffM | and if it dosn't exist, it won't compile |
| 16:20.22 | blast007 | DTRemenak: no, it won't |
| 16:20.30 | BulletCatcher | I can rewrite TimeLimit.cpp to use sscanf() rather than regex, and we can solve the bigger problem later. |
| 16:20.34 | blast007 | if you don't have config.h, you get "#define regex_t void" |
| 16:20.46 | blast007 | it checks for HAVE_REGEX_H |
| 16:20.50 | JeffM | I'll add "plugin lib issues" to the roadmap |
| 16:20.52 | blast007 | which is in config.h ;) |
| 16:21.01 | DTRemenak | ah, shucks :) |
| 16:21.22 | JeffM | BulletCatcher, that sounds like the safest thing to do now |
| 16:21.24 | DTRemenak | I'm sure this is not the last plugin that's going to have non-bzfs dependencies |
| 16:21.42 | JeffM | probably not, so we need to figure out what we want to do, just not today |
| 16:21.51 | DTRemenak | if you want them to be independent, then they'll need their own independent build systems |
| 16:22.13 | DTRemenak | if you want them to be integrated, then they won't be independent |
| 16:22.35 | CIA-5 | BZFlag: 03JeffM2501 07http://my.bzflag.org * r7795 10/w/Development_RoadMap: |
| 16:22.52 | JeffM | just desuport windows server and it'll be a lot simpler :) |
| 16:23.09 | JeffM | then you don't even need this API thingy |
| 16:23.13 | DTRemenak | mac has the same problem with shipping binaries |
| 16:23.31 | JeffM | the mac build system dosn't allow for independence |
| 16:23.35 | JeffM | since it's tied to the linux on |
| 16:23.45 | JeffM | only the windows build system supports building one off plugins IIRC |
| 16:24.00 | JeffM | with out hacking up the makefiles I mean |
| 16:24.06 | DTRemenak | only the windows build system supports building our own plugins externally from the rest of the source |
| 16:24.11 | DTRemenak | one off plugins are fine everywhere |
| 16:24.11 | JeffM | yes |
| 16:24.20 | JeffM | but nobody uses windows to host :) |
| 16:24.21 | DTRemenak | since they have their own build systems |
| 16:24.23 | JeffM | so it's somewhat moot |
| 16:25.20 | DTRemenak | that could be done for all our plugins, using configure chaining, just like we do for external libs |
| 16:25.26 | DTRemenak | it'd just make builds take forever |
| 16:25.42 | JeffM | don't they already take that long on linux? :) |
| 16:26.01 | DTRemenak | heh. autogen+configure takes longer than the actual build, yeah :) |
| 16:32.55 | mdskpr_ | doesn't bzfsHTTPAPI.h make plugin_utils/plugin_http.h worthless? |
| 16:33.14 | JeffM | that's still there? |
| 16:33.16 | mdskpr_ | yes |
| 16:33.24 | JeffM | I thought I removed that |
| 16:33.27 | JeffM | but yeah it does :) |
| 16:33.28 | ts | DTRemenak: Hmm, for me build takes longer than autogen+configure |
| 16:33.43 | ts | DTRemenak: while building it's idling a lot |
| 16:34.01 | DTRemenak | ts: I use make -j4 too, so maybe not fair :) |
| 16:34.07 | JeffM | do you have it do more then one at a time? |
| 16:34.22 | JeffM | yeah -j2 helps even on single core systems |
| 16:34.32 | blast007 | -j2 is the default afaik |
| 16:34.47 | mdskpr_ | JeffM: i'll remove it |
| 16:34.57 | JeffM | mdskpr_, you can't just remove the files |
| 16:35.26 | mdskpr_ | ik |
| 16:35.32 | JeffM | k |
| 16:35.43 | JeffM | you can pull the base 64 stuff too |
| 16:49.09 | BulletCatcher | Windows seems to have the strspn() function, so I will use that too. |
| 16:57.41 | CIA-5 | BZFlag: 03mudskipper * r22041 10/trunk/bzflag/plugins/plugin_utils/ (10 files): removed base64, plugin_HTTP, templates_readme.txt, and base64.txt from plugin_utils. |
| 17:12.37 | CIA-5 | BZFlag: 03bullet_catcher * r22042 10/trunk/bzflag/plugins/TimeLimit/TimeLimit.cpp: The regular expression library is not properly exposed to plugins, so instead use sscanf() and strspn() for TimeLimit parameter handling. |
| 17:12.54 | BulletCatcher | That needs testing, especially on Windows. |
| 17:14.40 | BulletCatcher | did test before committing |
| 17:14.58 | DTRemenak | tsk tsk |
| 17:15.01 | DTRemenak | ;) |
| 17:15.05 | BulletCatcher | :-p |
| 17:16.47 | BulletCatcher | I tested with: |
| 17:16.51 | BulletCatcher | src/bzfs/bzfs -dd -a 0 0 -password x -world misc/maps/flagbuffet.bzw -time 600 -timemanual -loadplugin plugins/TimeLimit/.libs/TimeLimit.so,15-30 |
| 17:17.01 | blast007 | I see your password! |
| 17:17.03 | blast007 | bwahahaha |
| 17:17.12 | BulletCatcher | also: 15,20,30 |
| 17:17.20 | blast007 | and without anything at all? |
| 17:17.24 | BulletCatcher | That too |
| 17:17.24 | blast007 | just TimeLimit |
| 17:17.25 | blast007 | k |
| 17:17.27 | DTRemenak | accidentally ran a public server with that password once |
| 17:18.22 | BulletCatcher | Using the password to become admin is the easiest way to get the privileges needed to use the /timelimit commands. |
| 17:18.22 | JeffM | it builds on windows |
| 17:18.26 | BulletCatcher | yay |
| 17:18.45 | DTRemenak | bah, JeffM can build a lot faster than I can :) |
| 17:18.56 | JeffM | cus I'm actualy at work ;) |
| 17:22.49 | trepan | something else to consider, stick bzfsAPI.h in its own directory; make it harder to do mistakes like #include "bzregex.h" |
| 17:33.20 | CIA-5 | BZFlag: 03bullet_catcher * r22043 10/trunk/bzflag/plugins/TimeLimit/TimeLimit.cpp: |
| 17:33.20 | CIA-5 | BZFlag: Ignore the TimeLimit command line arguments when a time range has trailing non-digit characters. |
| 17:33.20 | CIA-5 | BZFlag: This is now the same as it was when a regular expression was used. |
| 17:33.57 | BulletCatcher | That is for a case like 15-30abc . |
| 17:36.05 | BulletCatcher | or more realistically: 15-20,30 |
| 17:43.49 | *** join/#bzflag djlawler (~dlawler@static-173-59-115-18.phlapa.fios.verizon.net) |
| 17:44.04 | *** join/#bzflag bryjen (~bryjen@76.92.85.169) |
| 17:44.04 | *** mode/#bzflag [+v bryjen] by ChanServ |
| 17:45.10 | *** part/#bzflag djlawler (~dlawler@static-173-59-115-18.phlapa.fios.verizon.net) |
| 17:45.12 | *** join/#bzflag djlawler (~dlawler@static-173-59-115-18.phlapa.fios.verizon.net) |
| 17:49.54 | *** part/#bzflag djlawler (~dlawler@static-173-59-115-18.phlapa.fios.verizon.net) |
| 17:52.44 | *** part/#bzflag Upsetter (~er@i59F6CB75.versanet.de) |
| 18:08.36 | *** join/#bzflag BulletCatcher (~bc@bzflag/developer/BulletCatcher) |
| 18:20.42 | JeffM | so do we want to do an RC3 today? |
| 18:24.32 | BulletCatcher | Only if you want to be *very* careful. We haven't changed much. |
| 18:24.39 | BulletCatcher | I'll be AFK in a moment, so won't be able to help. |
| 18:26.25 | JeffM | then do you think we should just release tomorow? |
| 18:27.24 | BulletCatcher | I can't think of a reason not to. |
| 18:27.34 | mdskpr_ | why not tonight? :P |
| 18:27.41 | JeffM | will everyone be arround? |
| 18:29.17 | BulletCatcher | I expect to be back in 8 hours, and I'll be here all day tomorrow. |
| 18:29.25 | JeffM | I think tomorow is better |
| 18:29.42 | BulletCatcher | Tommorrow it is, then. |
| 18:29.45 | JeffM | most of us are more western hemi time so we'll have more time to fix antyhing that comes up |
| 18:30.07 | *** mode/#bzflag [+o JeffM] by ChanServ |
| 18:30.24 | *** topic/#bzflag by JeffM -> http://my.BZFlag.org || http://cia.vc/stats/project/BZFlag || http://my.BZFlag.org/w/Getting_Help || Channel Logs: http://ibot.rikers.org/%23bzflag/ || 2.4 release is tomorow, be scared || The PLAN http://my.bzflag.org/w/Development_RoadMap |
| 18:59.02 | *** join/#bzflag meeba (~lamer@c-71-196-238-53.hsd1.co.comcast.net) |
| 19:22.21 | Pimpinella | sorry that i can't help much atm, got too much work |
| 19:22.58 | JeffM | thats why tomorow is better :) |
| 20:53.35 | JeffM | blast is there a max size to the hex string we store in the list server? the one that holds the game type and players etc.... |
| 20:56.41 | *** join/#bzflag djlawler (~dlawler@static-173-59-115-18.phlapa.fios.verizon.net) |
| 20:56.54 | *** part/#bzflag djlawler (~dlawler@static-173-59-115-18.phlapa.fios.verizon.net) |
| 21:04.11 | *** join/#bzflag djlawler (~dlawler@static-173-59-115-18.phlapa.fios.verizon.net) |
| 21:16.33 | *** join/#bzflag delusional (~delusiona@unaffiliated/delusional) |
| 21:24.16 | *** part/#bzflag djlawler (~dlawler@static-173-59-115-18.phlapa.fios.verizon.net) |
| 21:31.08 | *** join/#bzflag sirquine (~quine@c-71-237-73-109.hsd1.co.comcast.net) |
| 21:38.38 | trepan | JeffM: include/Ping.h PingPacketHexPackedSize ? |
| 21:40.25 | JeffM | huh? |
| 21:40.53 | trepan | " <@JeffM> blast is there a max size to the hex string we ..." |
| 21:41.04 | JeffM | ahh ok |
| 21:41.13 | JeffM | well I was curious about the db |
| 21:49.28 | JeffM | cus it stores it as text |
| 21:49.31 | trepan | `gameinfo` varchar(73) NOT NULL default '' |
| 21:49.38 | JeffM | ahh thanks |
| 21:49.53 | JeffM | wonders how much data he can gzip into 73 bytes |
| 21:49.53 | trepan | (assuming i've got the right database / table :) |
| 21:50.19 | trepan | thinks that the answer may be ironic |
| 21:50.40 | Pimpi | if it's not long enough, make it longer |
| 21:51.39 | Pimpi | no? |
| 21:53.58 | trepan | zlib/compress.c/compressBound(sourceLen) { return sourceLen + (sourceLen >> 12) + (sourceLen >> 14) + (sourceLen >> 25) + 13; } |
| 21:54.59 | JeffM | yeah I'd have to encode it too to be a valid string |
| 21:58.51 | *** join/#bzflag KTL (~KTL@85.234.201.239) |
| 22:25.02 | blast007 | JeffM: it'd be kinda silly to gzip stuff into there |
| 22:25.11 | blast007 | better off just doing it right |
| 22:37.59 | JeffM | well the right way is for me to write my own list server and not try to use bzflag's :) |
| 23:21.12 | *** join/#bzflag hyperiox (~hyperiox@109-254.198-178.cust.bluewin.ch) |
| 23:37.20 | GvzEvxre | no need to be backwards compatible for new clients. just look at the useragent or similar and send longer/different strings if they are newer |
| 23:37.50 | GvzEvxre | it would be nice to include multiple ip addresses per server to allow dual stack servers for example. |
| 23:38.04 | GvzEvxre | ipv4/ipv6 replies |
| 23:38.13 | blast007 | something like JSON would be a better format to send |
| 23:38.37 | blast007 | or even XML, though I prefer JSON :) |
| 23:38.59 | GvzEvxre | blast007, either is a possibility as long as we check for new clients first. |
| 23:39.16 | blast007 | we already send the version string |
| 23:39.18 | blast007 | afaik |
| 23:39.34 | GvzEvxre | yes. and this is why. |
| 23:40.48 | GvzEvxre | the current list protocol is lightly modified from what the old C++ implementation sent. re-implemented as php. |
| 23:41.42 | GvzEvxre | hmm. might have been c actually. I don't recall. bzfls.c? something like that. |
| 23:41.50 | trepan | blast007: you can already request lua/json formats, just tweak the GET a bit |
| 23:41.58 | trepan | so it's already backwards compat |
| 23:42.17 | blast007 | trepan: the current JSON is more like CSV than what I'm used to for JSON.. |
| 23:42.21 | GvzEvxre | trepan, that only helps if we pay attention to the requested format, which we don't at present. |
| 23:42.41 | trepan | blast007: it's JSON, without the indenting - and more compact |
| 23:42.57 | trepan | i dumped the lua out in almost exactly the same format |
| 23:43.08 | blast007 | JSON supports named keys, I mean |
| 23:43.16 | trepan | knows that all too well |
| 23:43.19 | trepan | so does lua |
| 23:43.30 | trepan | but naming each entry is a waste of space |
| 23:44.10 | blast007 | easier to access the data though |
| 23:44.26 | trepan | and named columns is just as flexible (and in lua's case, does not forbid named keys as well) |
| 23:44.28 | blast007 | less processing steps, at lesat ;) |
| 23:44.59 | trepan | 3 lines inconvenience ... is that ... might it be ... the world's smallest violin? :) |
| 23:45.08 | GvzEvxre | so extend the db field size if needed, the old servers/clients won't use the extra bytes. only send new clients json, and change the client parser. |
| 23:46.01 | trepan | or add a new field |
| 23:46.20 | trepan | ex: the server owner property is only available in the lua/json formats |
| 23:48.48 | GvzEvxre | nods. the space separated original format is unfortunate as it prevents adding new fields. |
| 23:49.28 | GvzEvxre | who/what uses the json/lua formats now? |
| 23:55.25 | trepan | has a lua script he uses to list servers and their owners :) |
| 23:56.07 | trepan | i added json at the some time as lua because 1. it was really easy 2. i figured it might be useful for a web script someday |
| 23:56.45 | trepan | blast007: fwiw, we could change the current lua/json to lua_compact/json_compact, and use full key/value pairs for lua/json |
| 23:57.24 | trepan | iirc, the lua compact format is about 47KB vs 42 KB for the space separate list |
| 23:57.27 | blast007 | I'll have to think about the pro/cons of that.. might not be worth it |
| 23:57.44 | trepan | blast007: very easy switch, better to do it now then later |
| 23:58.07 | blast007 | well, I also don't like the hex string ;) |
| 23:58.13 | trepan | ditto |
| 23:58.42 | blast007 | any idea if our cURL setup supports gzip? |
| 23:59.01 | trepan | dunno, but we do have the zlib available |
| 23:59.05 | blast007 | yeah |
