Thursday, 2010-06-17

*** tpb has joined #tp00:00
*** ChanServ sets mode: +o tpb00:00
*** epyon-sakura has joined #tp00:35
*** epyon-sakura has quit IRC01:23
*** nash has joined #tp01:31
*** epyon-sakura has joined #tp01:42
*** verhoevenv_ has quit IRC01:56
*** Erroneous has quit IRC02:12
*** epyon-sakura has quit IRC02:40
*** epyon has quit IRC02:40
*** epyon-okami has quit IRC02:40
*** epyon-sakura has joined #tp02:54
*** epyon-kitsune has joined #tp02:59
*** Epyon has joined #tp03:01
epyon-kitsunellnz: is there any reason we need the set of active connections?03:09
epyon-kitsunemaintaining a set of them will violate thread separation03:09
llnzit should be possible to pass that down to asio03:11
epyon-kitsunegreat :)03:11
llnzyou will need connection objects though03:11
epyon-kitsuneactually not asio, but shared pointers03:12
epyon-kitsunethey allow objects to live because asio will be holding them in the async loop03:12
epyon-kitsuneyes I know03:12
llnzok, can we get some design text/diagrams?03:12
epyon-kitsunebut I won't need socket objects, nor distinction between http etc diagrams03:13
epyon-kitsuneerr classes03:13
epyon-kitsuneI'll try to do a diagram for the next blog update03:13
*** glew has quit IRC03:13
llnzcol03:13
llnzcool, even03:13
epyon-kitsuneThere'll be a single Server class that'll be responsible for the threaded io_service launch, and will be creating 5 acceptor classes03:14
epyon-kitsunethese will have a callback that will pass the accepted sockets to the server which will create a proper (Admin or Player) connection class that will be thrown into the asio proactor again03:15
llnzok03:15
llnzbtw, it's probably double the number of listen sockets open03:16
llnzas ipv6 sockets no longer bind to IPv4 as well03:16
llnz(at least on debian)03:16
epyon-kitsunethe server will have an explicit state enum that will tell if we're processing a turn or running it03:16
llnzcool03:17
epyon-kitsunellnz: yeah, that also means will need two separate acceptor classes03:17
llnzreally?03:17
epyon-kitsune(or maybe I'll use a template argument for that)03:17
llnzthe only difference is the address class03:17
epyon-kitsuneconsidering that the acceptor class is very slim, I guess I could template it based on the IP protocol03:18
llnzand don't forget that it can be set to an explict bind address as well03:18
epyon-kitsunehmm, correction :03:20
epyon-kitsuneseems that the bind method I use is independent of IP protocol03:20
epyon-kitsuneit resolves the passed address and chooses protocol based on that03:21
llnzoh, that was the other thing, i would like to see code03:21
epyon-kitsunellnz: where should I push it?03:22
llnzthe refactor branch on git.tp.net if you want03:23
epyon-kitsunealso, just to be clear -- I'm building the ASIO server besides the existing implementation03:23
llnza branch on github if you have to03:23
epyon-kitsune(it seems easier)03:23
llnzcool03:23
*** epyon-sakura has quit IRC04:05
*** StupidIncarnate has joined #tp05:16
* llnz wanders off05:20
llnzlater all05:20
*** llnz has quit IRC05:20
*** tansell-laptop has quit IRC05:31
*** bisc has joined #tp06:54
bisctansell: ping07:38
tansellbisc, pong07:39
bisctansell: would be glad to get your reviews on open issues (again, except 59001and 54002)07:40
tansellbisc, looking at 5300107:47
tansellfreeze/thaw doesn't effect window size/position07:47
tansellit effects the calculation of dynamic sizers07:47
bisctansell: oh, well. Then I'll delete these calls. Any way to make window calculate its size, but not display immediately on screen?07:49
tansellbisc, not to my knowledge07:49
bisctansell: ok. But I personally didn't notice this window appearing in wrong position before movement. Do you think calling Show before Move is unacceptable?07:50
tansellbisc, is the window a fixed size?07:51
tansell(or could it be a fixed size?)07:51
bisctansell: no, it's not fixed. Its size greatly depends on 1) length of available filters 2) whether some filters are selected. And checklist boxes dynamically resize. We can try setting fixed size, but then we risk getting some filters labels not fit in.07:53
tansellbisc, shouldn't there be a scroll bar if there are too many filters?07:56
bisctansell: not too many, but too long filters. Currently no, the window is resized. Do you think I should apply horizontal scrollbar there and set fixed size?07:57
tansellbisc, you sure?08:00
tansellI can't see code which does the resizing?08:00
bisctansell:08:01
bisctansell: aren't they resized because of sizers?08:02
bisctansell: let be check this08:02
tansellokay08:02
bisctansell: I have found out these facts: 1) no fixed size for filter frame (and its components) is set in either code or xrc. 2) both horizontal and vertical sizes are currently by lenth and number of filters.08:13
bisctansell: so, there are 2 ways to handle it : 1) set fixed size, call move before show 2) set no fixed size, call move after show08:14
bisctansell: ...are currently _determined_ by lenth and number of filters.08:15
bisctansell: and one more interesting thing: calling move after show makes the filter frame align close to the screen border, not going over it (while theoretically it should).08:19
bisctansell: so, if we have this Messages panel in another part of screen -- we'll have to check whether this frame is out-of-screen. But calling Move after Show solves it?08:20
bisctansell: misprint. not ?, but !08:20
tansellbisc, hrm....08:36
tansellbisc, we don't really want the window to go off the screen do we?08:37
bisctansell: of course we don't.08:37
tansellso I must mis-understand what you are saying then08:37
bisctansell: yeah, sorry, let me refolmulate.08:38
bisctansell: two global options to organize the window: 1) move before show (with explicitly set coordinates) 2) move after show (with coordinates taken from window). Currently we have the second and you don't like it. I say that there is problem: filter frame can be off screen. If taking the first option, we will need to solve this problem with explicit checking whether the window is out of screen.08:42
bisctansell: and if we leave the second option, this is done automatically.08:42
bisctansell: don't know why it is done automatically, but the frame doesn't want to go off screen if we call Move when the window is shown.08:46
bisctansell: have my words become more clear?08:51
*** null_000 has joined #tp09:02
null_000hi all09:03
*** nash has quit IRC09:10
*** StupidIncarnate has quit IRC09:29
*** JLP has quit IRC10:28
*** JLP has joined #tp10:59
*** JLP has joined #tp10:59
*** null_000 has quit IRC11:28
*** glew has joined #tp12:16
ezodglew: hey12:49
*** verhoevenv has joined #tp12:50
ezodglew: still no github forks?12:54
*** bisc has quit IRC13:27
ezodso how is everyone's GSoC projects going?13:28
ezoddon't work in a bubble, let's hear about it :)13:28
alanpbut bubbles are so much fun13:29
verhoevenvLike this? http://www.youtube.com/watch?v=ART2V2ICxZI13:44
tpbTitle: YouTube - My Bubbles (at www.youtube.com)13:44
*** JLP has quit IRC14:12
*** JLP has joined #tp14:12
*** JLP has joined #tp14:12
*** Agon has joined #tp15:54
Agonhello16:01
AgonIs the velocity of an object in tp03 used to describe the direction in which eg. a fleet moves or can it be used for that?16:04
*** Erroneous has joined #tp17:23
*** null_000 has joined #tp17:57
*** null_000 has quit IRC18:17
*** llnz has joined #tp20:10
llnzmorning all20:11
ezodhey llnz20:12
llnzhi ezod20:12
Agonhi llnz20:14
llnzhi Agon20:14
llnzAgon: how are you going?20:14
Agonllnz: Good, I understand now how to get ObjectParameters. But I have a question about what ObjectParameter::getName() returns?20:17
AgonI want to identify what ObjectParameter I have.20:17
llnzok, first a couple of things:20:19
llnzdid you know the libtpprotocol had doxygen documentation?20:19
llnzdid you know that documentation is online20:19
llnz?20:19
Agonoh ok, thanks, found it20:21
Agonwell, "Gets the name of this ObjectParameter. " is not really helpful. Is it the class name or some protocol name?20:24
*** null_000 has joined #tp20:28
null_000hi all20:29
llnzit's the name the server has said that the particular object parameter is called20:29
llnzso it could be called "Position" and be a Position3dObjectParam20:29
llnzhi null_00020:29
null_000llnz: hi20:29
null_000llnz: did my answers help?20:31
llnzjust reading it now20:31
null_000ok20:33
*** bisc has joined #tp20:51
llnznull_000: reply sent20:55
null_000reply read20:56
null_000if you need any more info just ping me20:56
llnzyeah, will do20:57
null_000llnz: thanks for taking your time to check this out21:00
llnzfound it21:11
llnzobjectview doesn't get updated21:11
llnzwill fix it tonight21:11
*** StupidIncarnate has joined #tp21:24
null_000great!21:43
StupidIncarnategreat indeed?21:43
alanpnice22:36
StupidIncarnate does anyone know how to get the name of an object type if you have the typeid?23:01
StupidIncarnatetried this, but wants more arguments objdesc = objects.ObjectDesc()[obj.subtype]23:01
*** Vadtec has quit IRC23:10
*** Vadtec has joined #tp23:13
*** Epyon has quit IRC23:16
*** Epyon has joined #tp23:16
*** bisc has quit IRC23:22
StupidIncarnatebah, nvm. Mispelled23:25
*** null_000 has quit IRC23:51
*** alanp has quit IRC23:53

Generated by irclog2html.py 2.13.1 by Marius Gedminas - find it at mg.pov.lt!