| 00:00.08 | JeffM | you define a static variable of your ojbect type and init it once. |
| 00:00.14 | JeffM | there are many ways to do singletons |
| 00:01.51 | JeffM | one method is to just add a static variable to your class definition, since classes themselves define something namespace like |
| 00:02.03 | JeffM | so CURLManager::Instance or something |
| 00:03.03 | blast007 | using curl directly, not as part of cURLManager |
| 00:03.30 | JeffM | ahh then you'd just make a new namespace in a header, then add a static variable of CURL handle, or whatever it uses |
| 00:03.35 | JeffM | define it extern |
| 00:03.45 | blast007 | https://github.com/BZFlag-Dev/bzflag/pull/59 |
| 00:03.51 | JeffM | then in some Cpp file, init it and call it static |
| 00:03.55 | blast007 | using it inside the TextUtils namespace |
| 00:04.32 | blast007 | (so needing to define it within there and only use it within there) |
| 00:04.48 | JeffM | oh, so you don't want him to call curl_easy_init every time? |
| 00:05.48 | JeffM | just move CURL *curl = curl_easy_init(); above the funciton calls and make it static. |
| 00:05.55 | JeffM | then take it out of everywhere else |
| 00:06.09 | JeffM | then you'll get a single CURL ptr once the first time it's used |
| 01:27.03 | *** join/#bzflag I_Died_Once (~I_Died_On@unaffiliated/idiedonce/x-1828535) |
| 01:30.46 | *** join/#bzflag I_Died_Once_ (~I_Died_On@unaffiliated/idiedonce/x-1828535) |
| 01:34.49 | *** join/#bzflag I_Died_Once (~I_Died_On@unaffiliated/idiedonce/x-1828535) |
| 01:36.46 | *** join/#bzflag alpha1-2 (~alpha1@host10.200-43-202.telecom.net.ar) |
| 01:54.27 | *** mode/#bzflag [+v alpha1-2] by blast007 |
| 01:55.01 | blast007 | < alpha1-2> why can it be that flag is not passed when I drop it on an inclined surface? |
| 01:55.35 | blast007 | so what happens when it drops? does it return to the team's base? |
| 01:55.53 | alpha1-2 | no, it just falls down |
| 01:56.06 | alpha1-2 | through surface to ground |
| 01:57.02 | alpha1-2 | sometimes if I jump over some distance from surface it passes |
| 01:57.16 | alpha1-2 | (broken bridges in Castle Ruins) |
| 01:59.26 | alpha1-2 | cods is researching about server treating empty/inside spaces in buildings as allowable places to spawn; maybe bridge is too wide |
| 01:59.57 | blast007 | yeah, cound be inside points being the problem, hard to say though |
| 02:00.17 | JeffM | yay, teleporters can be drawn.. :) now they just need IDs. https://usercontent.irccloud-cdn.com/file/jmSexDW6/ |
| 02:00.21 | alpha1-2 | okay |
| 02:02.45 | blast007 | for instance, on earlier versions of Louman's Metropolis map, flags would spawn inside of buildings |
| 02:03.02 | blast007 | inside points fixed that |
| 02:08.18 | alpha1-2 | is reading about that |
| 02:15.46 | alpha1-2 | oh, the glitches, like shots passing through walls/bases in HIX are also related to this issue or they are a different one? |
| 02:35.14 | alpha1-2 | cods https://wiki.bzflag.org/Mesh#Inside_Points |
| 02:39.12 | alpha1-2 | a simple way which is worth it to try maybe: "try to fix it by add "inside 0 0 0" to the meshes" ("Sketch-Up Map Problems" in Forums) |
| 03:29.29 | allejo | took flash's suggestion and cURL's functions can work with null where a CURL object would be needed so that'd get rid of the curl_easy_escape() overhead |
| 03:30.22 | allejo | and since cURLManager already has a CURL object in it and is doing its own thing, I don't think a singleton would be needed for TextUtils, right? |
| 03:36.41 | allejo | however, he did link to a SO post, don't know if that'd affect us or not |
| 03:36.45 | allejo | http://stackoverflow.com/questions/14867404/curl-easy-escape-not-working-from-c-dll |
| 04:43.06 | The_Hand_of_Mi-2 | orly? |
| 06:38.57 | allejo | jeffm, the bottom of bases are tiled for your unity package |
| 07:53.06 | *** join/#bzflag Shuist (~Shuist@ppp203-122-213-220.static.internode.on.net) |
| 08:13.04 | cods | regarding what Alpha told above, I already explained that it was because of the algorithm that build meshes. So nothing wrong with bzflag itself. |
| 08:13.20 | cods | (and I do add inside points) |
| 08:17.08 | cods | to clarify, I build maps with Python using a CSG library which is not perfect (which will be replaced with my own implementation soon) and sometimes produces meshes that are non-manifold |
| 10:13.50 | blast007 | allejo: both the escape and unescape cURL methods make use of the CURL* object |
| 14:45.10 | allejo | blast007: https://curl.haxx.se/mail/lib-2006-04/0093.html |
| 14:46.08 | blast007 | hmm, okay |
| 14:47.17 | allejo | :D |
| 14:48.17 | blast007 | looks like the older curl_escape/curl_unescape does call the easy versions with a NULL there |
| 14:50.23 | blast007 | I'm confused though... curl_easy_escape calls Curl_convert_to_network with that CURL* handle |
| 14:55.28 | blast007 | and Curl_convert_to_network takes a 'struct SessionHandle *data' in that place, and then does 'if(data->set.convtonetwork) {' |
| 14:55.48 | allejo | beat me to it :p |
| 14:56.07 | blast007 | so isn't that bad if data is NULL? |
| 15:00.08 | blast007 | maybe I'm just thinking of PHP instead of C/C++ where calling something like "if ($data->thing['array'])" would throw an error/warning if $data wasn't an object |
| 15:02.34 | allejo | what's the constructor for SessionHandle look like? |
| 15:03.07 | blast007 | no idea, can't find where it's defined aside from stuff like "struct SessionHandler;" from github's web interface |
| 15:03.21 | blast007 | don't have a clicky clicky IDE here |
| 15:04.27 | allejo | but those are the best type! |
| 15:09.10 | allejo | hmm https://github.com/curl/curl/blob/master/include/curl/curl.h#L99 and https://github.com/curl/curl/commit/434f8d0389f2969b393ff81ead713b7600502f27 |
| 15:09.31 | allejo | but I can't seem to find that in our deps |
| 15:10.50 | blast007 | "our deps" meaning the bzflag-dependencies repo? |
| 15:10.56 | allejo | yea |
| 15:13.02 | blast007 | the one in our repo is 7.43.0, that change was in 7.50.* |
| 15:14.13 | blast007 | which apparently I should update ;) |
| 15:14.36 | blast007 | "curl escape and unescape integer overflows" |
| 15:14.55 | allejo | right, but in 7.50.* SessionHandle (now, Curl_easy) is a typedef of CURL. so, trying to see if that applied to prior versions too |
| 15:15.08 | allejo | lol yay overflows! |
| 15:16.11 | blast007 | which is apparently fixed in debian oldstable, but *not* in stable, testing, or unstable... |
| 15:18.40 | blast007 | might not really apply to us though since it requires a string length of 0xffffffff |
| 15:18.59 | blast007 | that'd exceed our other buffer lengths :) |
| 15:20.00 | allejo | oh we better be careful with those long strings |
| 15:38.44 | JeffM | allejo: thanks, I fixed it. |
| 15:46.29 | *** join/#bzflag Zehra (~Zehra@unaffiliated/zehra) |
| 15:54.53 | JeffM | what's the prefered way to do TP links these days? by face ID or by name? |
| 15:55.25 | blast007 | for an editor, probably by ID |
| 15:55.52 | blast007 | unless you autogen names and use those |
| 15:55.57 | JeffM | I can autogen names |
| 15:56.07 | JeffM | ID is hard, it's based on the order in the file |
| 15:56.19 | JeffM | so add a TP and the IDs shift |
| 15:56.34 | JeffM | I can also also ensure there is a name |
| 15:56.50 | JeffM | I'm generating a GUID for each object, so I have a unique name I can use if needed |
| 16:26.56 | *** join/#bzflag [ddmp] (~dennis@unaffiliated/dmp/x-546784) |
| 16:51.26 | *** join/#bzflag Zehra (~Zehra@unaffiliated/zehra) |
| 17:10.41 | *** join/#bzflag Zehra (~Zehra@2601:58b:c401:df8:551c:1c1c:f08f:bc03) |
| 17:10.42 | *** join/#bzflag Zehra (~Zehra@unaffiliated/zehra) |
| 17:12.27 | *** join/#bzflag Zehra (~Zehra@unaffiliated/zehra) |
| 17:55.52 | *** join/#bzflag Zehra (~Zehra@unaffiliated/zehra) |
| 18:15.57 | *** join/#bzflag Zehra (~Zehra@unaffiliated/zehra) |
| 18:26.42 | *** join/#bzflag Zehra (~Zehra@unaffiliated/zehra) |
| 19:34.34 | *** join/#bzflag Zehra (~Zehra@unaffiliated/zehra) |
| 20:08.02 | JeffM | what was the thing with porter links that bzedit did wrong? |
| 20:09.22 | JeffM | wiki says it should be name:f/:b or * |
| 20:14.51 | blast007 | having 'name something' lines in the teleporter defintion breaks index based links |
| 20:48.47 | *** join/#bzflag Shuist (~Shuist@ppp203-122-213-220.static.internode.on.net) |
| 21:41.21 | JeffM | so if I have names I must use names, if i use face indexes I must use face indexes? |
| 21:41.26 | JeffM | and not names |
| 21:42.39 | JeffM | I just want to pick one way to use, and have the IO lib fix maps that use the other way. so converting face indexes to names on load seems to be the easy way, then I just make sure everything has a name. |
| 21:53.47 | blast007 | the teleporter syntax is weird. there's a name after the 'teleporter' keyword, as well as a 'name' attribute under the teleporter keyword |
| 21:55.08 | blast007 | docs say that the one after the 'teleport' keyword (as in, 'teleporter something\nposition 0 0 0\nsize 5 1 5\nend') is used for linking |
| 22:15.23 | JeffM | ohh odd |
| 22:15.42 | JeffM | that'd be one of the rare multi-word objects then |
| 22:15.50 | JeffM | I'll have to modify yon parser |
| 22:22.33 | JeffM | so should I disallow the "name" attribute? have it overide the name after the keyword? |
| 22:30.28 | blast007 | no idea, I'm not clear on how it works |
| 22:30.46 | JeffM | hates reading that parsing code |
| 22:33.10 | JeffM | the output code that writes porters only ever puts the name after the word "teleporter", it does not seem too use a "name" attribute |
| 22:34.29 | JeffM | I'll parse a name attribute and have it set the name field only if one isn't specified after the keyword |
| 22:34.36 | JeffM | that way the parser will fix busted maps |
| 22:37.44 | JeffM | by output code I mean the stuff from -saveworld |
| 22:38.02 | JeffM | that's probably the best thing to base it on, how the client saves it's own maps |
| 22:49.48 | *** join/#bzflag Zehra (~Zehra@unaffiliated/zehra) |