| 00:02.46 | starseeker | d-lo: are you yawning at subversion? :-P |
| 00:03.41 | d-lo | starseeker: nah, just been up since 0400 and Im looking to burn the midnight oil tonight :/ |
| 00:03.49 | starseeker | O.o |
| 00:03.53 | starseeker | eeep - how come? |
| 00:04.10 | d-lo | GS deadline :) |
| 00:04.18 | starseeker | ah, yes :-P |
| 00:04.24 | d-lo | I kinda freek out when I lose 6 months |
| 00:04.38 | starseeker | will take a stab at building that once he gets Qt set up... |
| 00:04.52 | d-lo | off my deadline :) |
| 00:05.01 | starseeker | nods |
| 00:05.05 | starseeker | that'll do it |
| 00:06.46 | starseeker | I'll be putting in couple hours myself here - gonna have to run some errands over the next couple days |
| 00:07.56 | ``Erik | when is this supposed deadline, again? |
| 00:08.32 | d-lo | Thursday :) |
| 00:09.06 | ``Erik | like, may 19, 2011? :D |
| 00:10.44 | ``Erik | well, good luck on that O.o |
| 00:14.26 | starseeker | where's the friggin Qt source code for mac? |
| 00:17.36 | starseeker | ah |
| 00:18.34 | starseeker | d-lo: is it any part of the requirements to have anything working on Windows? |
| 00:18.46 | louipc | starseeker: ftp://ftp.qt.nokia.com/qt/source/ ? |
| 00:19.01 | d-lo | not really no, although I have been trying to keep it compiling on winders and nix |
| 00:19.04 | starseeker | well, trolltech was the link I found, but that might work too :-) |
| 00:19.25 | starseeker | d-lo: hmm. Well, unfortunately my stuff is gonna be pretty much untested under Windows |
| 00:19.46 | d-lo | thats fine. |
| 00:20.05 | d-lo | we can always hammer it into cross platform later. |
| 00:22.55 | starseeker | d-lo: is 4.7 too new? |
| 00:23.04 | d-lo | dunno |
| 00:23.24 | d-lo | I am linking against 4.6.2 and haven't looked at the diff between 4.6.2 and 4.7.0 |
| 00:23.31 | d-lo | prolly nothing important |
| 00:23.53 | starseeker | tries it... |
| 00:25.24 | d-lo | drumroll |
| 00:25.26 | starseeker | um... how long does Qt take to build? |
| 00:25.36 | d-lo | got a snickers? |
| 00:26.06 | starseeker | sighs - figures |
| 00:26.24 | d-lo | took 8+ hours on a single core on winders |
| 00:26.31 | d-lo | took about 2 hours on multicore winders |
| 00:26.33 | starseeker | O.o |
| 00:26.41 | d-lo | 45 mins on 4 core *nix |
| 00:27.02 | d-lo | that was the full up build |
| 00:27.06 | d-lo | apps, libs, etc |
| 00:27.14 | starseeker | ``Erik: what say we stick Qt into the default BRL-CAD build? :-P |
| 00:27.46 | d-lo | muwahahahaha |
| 00:29.07 | starseeker | d-lo: have you tried this on the Mac? |
| 00:29.33 | d-lo | negative ghost rider |
| 00:29.57 | starseeker | goes with the no-framework config in the hopes it will work in a non-system dir... |
| 00:30.16 | starseeker | may mess with FindQt though... |
| 00:32.51 | starseeker | wonders why Apple doesn't just include Qt by default... |
| 00:32.52 | ``Erik | cool, then I can come in, update, run 'make', and go back home |
| 00:32.53 | ``Erik | O.o |
| 00:32.57 | ``Erik | cuz qt sucks? |
| 00:33.14 | starseeker | that doesn't appear to be the concensus |
| 00:33.22 | d-lo | f that ``Erik , just ssh in and start the build :) |
| 00:36.19 | ``Erik | heh, cron, dude |
| 00:36.55 | starseeker | ``Erik: careful, don't replace yourself with a very small shell script ;-) |
| 00:38.41 | ``Erik | whaheh |
| 00:39.18 | ``Erik | what's the hardest part about writing an rms simulator on a pdp8? |
| 00:39.29 | ``Erik | figuring out what to do with the other 3k of ram :D |
| 00:39.32 | starseeker | heh |
| 00:39.40 | ``Erik | (was a 4k machine) |
| 00:42.46 | d-lo | you could store a stupid-small graphic with 3k :) |
| 00:42.52 | starseeker | what's rms up to these days? |
| 00:43.16 | d-lo | married, three kids |
| 00:43.22 | ``Erik | http://www.netfunny.com/rhf/jokes/88q3/2104.6.html |
| 00:43.35 | starseeker | d-lo: riiiight, how about something plausible? |
| 00:43.37 | ``Erik | he was just in europe interrupt politicians speeches fighting software patents, iirc |
| 00:44.10 | ``Erik | think it was on smacksnot |
| 00:44.15 | ``Erik | http://www.itnews.com.au/News/232825,stallman-crashes-european-patent-session.aspx |
| 00:45.37 | starseeker | I'd say we should get Colbert to do a bit on software patents, except I'd guess that most of Congress wouldn't realize he was "in character" |
| 00:46.38 | ``Erik | the illegal immigrant worker bit was good... those congress critters really need their asses kicked good and hard |
| 00:47.01 | d-lo | okay, gunna unplug for a bit. peace all! |
| 00:47.21 | ``Erik | hasta manana, d-lo |
| 01:11.52 | starseeker | ah, excellent |
| 01:14.41 | CIA-2 | BRL-CAD: 03starseeker * r40718 10/rt^3/trunk/src/libPkgCpp/CMakeLists.txt: looks like libpkgcpp needs libbu as well |
| 01:20.47 | starseeker | d-lo: I can't compmile src/libNet/Portal.cxx is it working for you? pkg_switch and pkg_conn seem to be the issues - incorrect initialization on line 41 and a non-existant member of pkg_conn called pkc_user_data |
| 01:22.15 | starseeker | as near as I can tell, both of these complaints are quite valid, at least according to the data structures in pkg.h. |
| 01:24.20 | starseeker | am I missing something? a grep for pkc_user_data anywhere in rt3 or BRL-CAD comes up empty, except for Portal.cxx |
| 01:25.14 | ``Erik | pkg_user_data mebbe? |
| 01:25.31 | ``Erik | dave added a userdata symbol to libpkg last week iirc |
| 01:25.36 | starseeker | that's what I thought, but that's not in pkg.h either |
| 01:25.42 | starseeker | updates |
| 01:25.49 | starseeker | that could be why |
| 01:26.19 | ``Erik | huh, he called it pks_user_data in one struct and pkc_user_data in another |
| 01:26.35 | ``Erik | or, it is called |
| 01:27.06 | ``Erik | *shrug* updating should do it, though |
| 01:27.18 | starseeker | nods |
| 01:27.21 | starseeker | trying now |
| 01:28.53 | starseeker | d-lo: (when you get back on) do you want me to wire the building of the subversion libs into rt^3? |
| 01:46.21 | starseeker | yeah, got by libNet |
| 01:53.26 | ``Erik | heh, facebook convo between gollum and smeagol http://www.collegehumor.com/picture:1944771 |
| 01:59.06 | CIA-2 | BRL-CAD: 03starseeker * r40719 10/rt^3/trunk/cmake/FindBRLCAD.cmake: If we're only getting the header dirs from this mechanism, make sure we get what we need. |
| 02:02.05 | CIA-2 | BRL-CAD: 03starseeker * r40720 10/rt^3/trunk/include/brlcad/Combination.h: Dave, please check me here - I had to make m_tree public instead of private to compile the Combination.cpp file in the coreinterface, but I don't know if that's a Bad Thing. |
| 02:03.36 | starseeker | d-lo: Any "make test" kind of rule to fire off the test framework? |
| 02:05.43 | starseeker | has it all compiled now - yay! |
| 02:09.34 | starseeker | hey, cool - a BSD licensed alternative to GNU screen! http://tmux.sourceforge.net/ |
| 02:09.58 | starseeker | ponders evil thoughts about multiplexed mged terminal windows... |
| 02:18.06 | starseeker | alright, time to go |
| 05:34.27 | *** join/#brlcad merzo (~merzo@88.119.128.61) | |
| 08:24.20 | *** join/#brlcad d_rossberg (~rossberg@BZ.BZFLAG.BZ) | |
| 08:27.21 | d_rossberg | starseeker: it is a bad thing, what was the error message when compiling the Combination class with m_tree protected? |
| 08:32.36 | CIA-2 | BRL-CAD: 03d_rossberg * r40721 10/rt^3/trunk/cmake/FindBRLCAD.cmake: probable only the /bin at the end of the path should be replaced |
| 08:42.01 | starseeker | d_rossberg: I don't recall specifically, something about m_tree being protected (which is why I tried moving it) |
| 08:42.13 | starseeker | was compiling on OSX, if it matters |
| 08:45.57 | d-lo | starseeker: Sure, go ahead and wire in svn. |
| 08:46.41 | starseeker | d-lo: k |
| 08:47.27 | d-lo | oh and the include/brlcad and src/coreInerface stuff is all d_rossberg's baby :) |
| 08:48.28 | d-lo | fwiw I am still working out bugs, code inconsistencies, etc. the whole pks and pkc thing is just one discrepancy ;) |
| 08:50.27 | d-lo | so, starseeker you still awake or just getting up? |
| 08:55.32 | d-lo | transforms and rolls out |
| 08:56.47 | starseeker | d-lo: woke up for some reason |
| 08:58.45 | d-lo | woot! Just reserved Ironman two at the redbox :) |
| 08:58.54 | d-lo | okay, now Im leaving....fer reals |
| 08:59.04 | starseeker | hehe - and Blockbuster takes another hit |
| 09:07.23 | CIA-2 | BRL-CAD: 03starseeker * r40722 10/rt^3/trunk/include/brlcad/Combination.h: d_rossberg, author of the file, indicates the m_tree should be protected - try to figure out what's going on. |
| 09:09.38 | d_rossberg | starseeker: i'm currently trying to reproduce your problem in OSX |
| 09:10.12 | d_rossberg | Mac OS X version 10.5.8 |
| 09:10.15 | CIA-2 | BRL-CAD: 03starseeker * r40723 10/rt^3/trunk/cmake/FindBRLCAD.cmake: Replacing the last bin isn't enough if there's a list of paths - also replace '/bin:' patterns |
| 09:10.41 | starseeker | d_rossberg: I'm not at my Mac right now - it'll be a few hours before I can re-generate it, but if you can't reproduce it there I'll be sure to do so and post the build log |
| 09:12.05 | starseeker | hmm, getting something different on gentoo: |
| 09:12.06 | starseeker | /home/cyapp/cadtoplevel/brlcad/rt3/tests/libNet/libNetTest.cxx: In function âint main(int, char**)â: |
| 09:12.10 | starseeker | /home/cyapp/cadtoplevel/brlcad/rt3/tests/libNet/libNetTest.cxx:101: error: âtransformâ is not a member of âstdâ |
| 09:13.16 | d_rossberg | i tested the core interface on windows and linux (debian squeeze) |
| 09:13.26 | d_rossberg | try make coreinterface |
| 09:16.32 | CIA-2 | BRL-CAD: 03starseeker * r40724 10/rt^3/trunk/tests/libNet/libNetTest.cxx: Apparently need this include to get std::transform on gentoo linux (probably has something to do with gcc versions) |
| 09:16.42 | starseeker | d_rossberg: yeah, coreinterface builds on gentoo just fine |
| 09:16.52 | starseeker | it was apparently OSX specific |
| 09:17.03 | d_rossberg | is out for lunch |
| 09:20.57 | *** join/#brlcad cosurgi (~cosurgi@atak.bl.pg.gda.pl) | |
| 09:21.01 | starseeker | notes the same librt linker warnings he was seeing earlier for BRL-CAD itself |
| 09:32.13 | starseeker | yep |
| 09:32.20 | starseeker | _BRLCAD_LIBRARY_rt:FILEPATH=/usr/lib64/librt.so |
| 09:32.24 | *** join/#brlcad mafm (~mafm@83.49.86.17) | |
| 09:32.56 | starseeker | that's incorrect, should be /usr/brlcad/librt.so |
| 09:33.40 | starseeker | and just for more fun this box also has /lib64/librt.so.1 and /lib32/librt.so.1 |
| 09:34.34 | starseeker | oh joy, glibc itself owns that one |
| 09:35.32 | starseeker | yeah, thought so - symlink into /usr/lib64 from /lib64 |
| 09:36.15 | starseeker | or rather, both are symlinking through to librt-2.11.2.so |
| 09:40.08 | starseeker | I think I get away with it in the BRL-CAD build because librt is actually a defined target and I'm not having to find it.... let's see, maybe it is wrong |
| 09:51.21 | CIA-2 | BRL-CAD: 03starseeker * r40725 10/rt^3/trunk/cmake/FindBRLCAD.cmake: Try to avoid accidently spotting glibc's librt - it won't help if there's been an overwriting of glibc's librt, but BRL-CAD's install tries to avoid that as a rule. |
| 10:08.41 | *** join/#brlcad merzo (~merzo@88.119.128.61) | |
| 10:37.26 | d-lo | starseeker: thanks for the #include <algorithm>, I was gonna get that when I got to work :) |
| 10:37.52 | d-lo | its not needed on RHEL at work, but it is at home on ubuntu. |
| 10:43.56 | *** join/#brlcad d_rossberg (~rossberg@BZ.BZFLAG.BZ) | |
| 10:44.12 | *** join/#brlcad starseeker (~starseeke@BZ.BZFLAG.BZ) | |
| 10:56.02 | d-lo | Okay, so a 'const' in a struct... how is that handled? once that const field is set, it cannot be overwritten? |
| 10:58.32 | d_rossberg | const char* ? |
| 10:59.06 | d-lo | "const struct pkg_switch *pkc_switch" |
| 10:59.17 | d-lo | from the pkg_conn struct in libpkg |
| 10:59.32 | d-lo | but this is more of a "im still learning C" type question |
| 11:02.28 | d_rossberg | this means that you can reassign the pointer stored in pkc_switch but you may not change the struct your variable points to |
| 11:03.12 | d-lo | awesome, thanks :) |
| 11:04.58 | d-lo | so this means that if a 'const struct pkg_switch *switch" is initially set to NULL, then nothing after that can change the assignment to the variable 'switch' from NULL? |
| 11:06.31 | brlcad | ahh, excellent history to catch up on .... |
| 11:06.55 | d_rossberg | no, you can reassign the variable, but this holds only an address (a pointer) |
| 11:07.13 | d_rossberg | what is at this pointer is cont |
| 11:08.04 | d_rossberg | (or better: "at this address") |
| 11:08.33 | d-lo | Hrm, a bit confused. What is constant? a) the address that the 'switch' pointer points to, or b) the data that resides in that block of memory? |
| 11:08.50 | d_rossberg | b) |
| 11:09.02 | d-lo | okie. |
| 11:09.15 | d-lo | I suppose that makes sense, heh |
| 11:10.38 | d-lo | brlcad: for some reason, libpkg isn't calling my 'springboard function'. I am investigating, but my guess is that it has something to do with my 'late' setting of the pkg_switch table in pkg_conn |
| 11:11.14 | brlcad | k |
| 11:11.35 | d-lo | I might be going crazy by the time you get here and need help. lol |
| 11:11.48 | brlcad | can't be something too complex, I can investigate the pkg side to make sure it's calling right |
| 11:12.28 | d-lo | well i ran the tpkg client server test and that works just fine, so that little mod I made to libpkg doesn't seem to be the issue |
| 11:13.09 | brlcad | they dont' use the new callback data though :) |
| 11:13.58 | d-lo | I know :) I used it to verify that libpkg was still calling the callback function provided.... just to narrow it down to my code. |
| 11:14.28 | brlcad | I wouldn't expect any of the existing code to fail given you just added a new parameter |
| 11:14.49 | brlcad | they'll have warnings on their struct decls being incomplete, but that should be it |
| 11:15.04 | d-lo | hehe, thats what I had hoped, but I am not confident in my C coding skills :) |
| 11:15.06 | brlcad | might break strict build, but trivial to fix |
| 11:15.47 | brlcad | yeah, I suspect it's just not passing the parameter or something similar |
| 11:16.27 | brlcad | has tux and hotel, starseeker, tick tock getting close... ! |
| 11:16.51 | d-lo | okay, so when I call pkg_open and pass in a NULL for the pkg_switch table, am I able to set the pkg_switch table at a later date? or will the const prevent me? |
| 11:17.17 | d-lo | by what d_rossberg said above, I'd say no, since I can exchange the pointer at will... |
| 11:17.59 | brlcad | you might be able to change it, but doesn't mean pkg is using the changed one ;) |
| 11:18.12 | brlcad | i can investigate that bit while you work on something else |
| 11:19.34 | d-lo | I see :) |
| 11:20.00 | d-lo | hrm, well this little bug is kinda holding me up from moving forward, so I am gonna keep plugging away at it. |
| 11:20.08 | brlcad | you committed to head, right? |
| 11:20.20 | d-lo | correct |
| 11:20.29 | brlcad | looks like what I said, it's just not getting passed |
| 11:21.05 | d-lo | 'what' is not getting passed? the pkg_switch table or the pkg_user_data pointer? |
| 11:21.46 | brlcad | pkg_user_data |
| 11:21.48 | brlcad | fixing |
| 11:22.44 | brlcad | hm |
| 11:22.55 | brlcad | now it's starting to come back to me (as I wake up) |
| 11:23.48 | d-lo | the issue that I am having is that my static function Portal::springboard() isn't being called at all. |
| 11:24.08 | brlcad | yeah, I take it back -- it looks right |
| 11:24.11 | d-lo | so somehow my pkg_switch[] isn't getting passed into the pkg_conn correctly |
| 11:28.16 | CIA-2 | BRL-CAD: 03brlcad * r40726 10/brlcad/trunk/src/libpkg/pkg.c: few sanity sets to NULL on the user data after we're done with it |
| 11:29.47 | CIA-2 | BRL-CAD: 03brlcad * r40727 10/brlcad/trunk/include/pkg.h: ws |
| 11:32.08 | brlcad | pkg is just stashing the switch pointer passed to it, so if the callback isn't being called, it's either not getting set or the struct is no longer valid by the time it's called |
| 11:32.38 | d-lo | I think I am on to where/why its not being called. |
| 11:32.52 | d-lo | thanks for the help :) I'll let you know if/what i find |
| 11:45.07 | CIA-2 | BRL-CAD: 03brlcad * r40728 10/brlcad/trunk/src/libpkg/tpkg.c: data is merely sunk so no need to leverage the new user_data field. init to null. |
| 11:45.42 | d-lo | bingo. Just verified that the dataload is being recv'd and is trying to be dispatched, but _pkg_dispatch cannot find a handler. |
| 11:45.49 | d-lo | ...now, the fix :/ |
| 11:46.28 | ``Erik | might be worth reading up on trampolines and thunks |
| 11:46.46 | d-lo | might be :) |
| 11:47.43 | d-lo | but I have a stupid "X needs a Y during init, Y needs a Z during init, but Z needs an X during init" thing goin' on :/ |
| 11:51.21 | d-lo | okay, if a pkg_conn struct has a const pkg_switch* pointer, and the pkg_switch struct's fields are NOT const, do I have the ability to modify the fields in a pkg_switch struct after I have set it in a pkg_conn ? |
| 11:51.26 | d-lo | ...did that even make sense, lol |
| 11:56.27 | d_rossberg | starseeker: i could reproduce the error, i've a work-around in my mind, however how can i update gcc on a mac? |
| 12:00.27 | brlcad | d-lo: it depends when/where you access the switch |
| 12:01.20 | brlcad | if you're trying to access it through the conn, then that's no good |
| 12:01.34 | brlcad | if you're accessing the version you set in the switch, you're fine |
| 12:01.42 | d-lo | kk |
| 12:01.46 | d-lo | thats the route I am heading |
| 12:01.48 | brlcad | I suspect you're trying the prior |
| 12:01.53 | d-lo | just got the callback working :) |
| 12:02.09 | d-lo | now just have to get the userdata set correctly |
| 12:02.24 | d-lo | thanks for the guidance/mentoring :) |
| 12:02.43 | d-lo | I'll get a handle on this stuff, eventually, I promise ;) |
| 12:04.10 | starseeker | d_rossberg: you need to upgrade XCode, IIRC |
| 12:04.27 | starseeker | d_rossberg: you're saying it's a gcc problem? |
| 12:08.33 | starseeker | waits for d-lo's commit with baited breath... |
| 12:08.55 | d-lo | hey now, none of that breathing stuff. |
| 12:09.08 | starseeker | what, am I over my air quota again? |
| 12:09.33 | brlcad | breaths heavily .. and slowly .. asking d-lo .. "are you done yet?" |
| 12:10.01 | starseeker | brlcad: sweet, I got notice my tux is ready too so will probably pick it up later today |
| 12:10.09 | d-lo | ooookay, this just got creepy :P |
| 12:11.44 | d-lo | Is it bad juju to cast something out of const-ness? |
| 12:12.35 | d-lo | I know it indicates an inherit design issue, but will the men in black come for me? |
| 12:13.50 | CIA-2 | BRL-CAD: 03davidloman * r40729 10/rt^3/trunk/tests/libNet/libNetTest.cxx: forgot to run an int thru QString::number() |
| 12:14.25 | starseeker | d-lo: I've done it on occasion |
| 12:15.04 | starseeker | typically when I need to feed a value from somewhere into a function that didn't const its own use of the variable |
| 12:15.40 | starseeker | (if it really DOES change it its a problem, but sometimes it's just that the target function didn't const where they could have) |
| 12:18.01 | CIA-2 | BRL-CAD: 03davidloman * r40730 10/rt^3/trunk/src/libJob/JobWorker.cxx: Quell some debug comments in JobWorker |
| 12:18.26 | starseeker | alrightie, I may as well head in - I got a couple more hours sleep :-P |
| 12:19.20 | starseeker | Ooo, closing on the three month for the sourceforge takeover request on the nurbs stuff |
| 12:19.32 | starseeker | good timing |
| 12:19.58 | d-lo | wedding present :) |
| 12:20.05 | CIA-2 | BRL-CAD: 03davidloman * r40731 10/rt^3/trunk/src/libJob/JobManager.cxx: Quell some more debug comments in JobManager |
| 12:20.22 | brlcad | d-lo: muahaha |
| 12:20.36 | brlcad | and yes, it's considered VERY bad ju-ju |
| 12:20.58 | starseeker | heh - "congrats, here's a bunch of work to do to bring this project back to life!" |
| 12:21.24 | brlcad | nifty! |
| 12:21.34 | starseeker | Oct. 9th I'll know |
| 12:21.34 | d-lo | Well if a Portal contains a pkg_conn, but the pkg_conn needs to contain a reference to the Portal object... hows that gonna work? :) |
| 12:21.51 | starseeker | or 10th if they're slow off the gun :-P |
| 12:22.43 | d-lo | brainstorms |
| 12:23.00 | d-lo | PortalProxy object as an intermediary? |
| 12:23.08 | starseeker | then I can start to flaunt my horrible lack of practical C++ skills - up til now most of my C++ stuff has been training... |
| 12:23.20 | starseeker | alrightie, driving |
| 12:23.36 | d-lo | Be safe. No Pat style driving |
| 12:23.41 | d-lo | *snicker* |
| 12:23.51 | starseeker | <snort> - I'm a very boring driver, as a rule |
| 12:24.06 | brlcad | wonders if he can shower and arrive before cliff |
| 12:24.17 | d-lo | ohshi... its a race! |
| 12:24.20 | starseeker | just put your top down on the car |
| 12:24.21 | d-lo | go go go |
| 12:24.29 | starseeker | shower while driving :-P |
| 12:24.34 | brlcad | hehe |
| 12:24.37 | brlcad | good idea |
| 12:24.41 | starseeker | or didi it finally stop raining? |
| 12:24.53 | d-lo | the winner will receive one(1) U.S. American dollar! |
| 12:24.53 | brlcad | nope, still going here |
| 12:25.22 | starseeker | d-lo: as long as it was minted in 1794, sounds good! :-P |
| 12:25.23 | brlcad | d-lo: it'll have to get passed into the callback |
| 12:25.32 | brlcad | like it should have been to start with really |
| 12:25.37 | brlcad | then it's a non-const |
| 12:26.06 | brlcad | particularly if you need mutable data |
| 12:26.28 | starseeker | d-lo: say, this dollar? http://coins.ha.com/common/view_item.php?Inventory_No=200240026 :-P |
| 12:26.35 | d-lo | yack more when you beat starseeker, err, get into the office? |
| 12:27.15 | d-lo | starseeker: sure. that one. Just tell them I said you could have it. =D |
| 12:27.24 | starseeker | hehe |
| 12:27.35 | brlcad | you can probably get the same result with some casting since it is *your* data, but de-consting is usually a "really bad thing to do" to be avoided at all costs unless it's impossible |
| 12:28.18 | d-lo | what I ment by 'de-const-ing' it was that since the pkg_conn needs a valid pkg_switch table when the pkg_conn is created |
| 12:28.48 | d-lo | I fed it a switch table that had all the data except a proper user data pointer, since the object i want to point at didn't exist yet |
| 12:29.11 | d-lo | once the pkg_conn was created and i used that conn to create a Portal, THEN i set the userdata to point to the Portal |
| 12:29.35 | d-lo | but getting at that userdata involved a de-const-ing maneuver |
| 12:29.50 | d-lo | well, getting at it to modify it anyways.... |
| 12:30.07 | starseeker | what about making a non-const copy and feeding that in? |
| 12:30.07 | d-lo | its the only way I can see to beat the Chicken/Egg senario |
| 12:30.41 | d-lo | its the pkg_conn that makes the pkg_switch table const |
| 12:30.46 | starseeker | mm |
| 12:32.05 | *** join/#brlcad Zaebos (~irc@pd95b7f5e.dip0.t-ipconnect.de) | |
| 12:34.57 | CIA-2 | BRL-CAD: 03d_rossberg * r40732 10/rt^3/trunk/include/brlcad/Combination.h: |
| 12:34.58 | CIA-2 | BRL-CAD: work-around a bug in Mac OS X's gcc |
| 12:34.58 | CIA-2 | BRL-CAD: one day I'll remove it and see what happens }:-) |
| 12:51.06 | brlcad | right, that was my earlier point -- you can modify the switch, but you can't modify the switch *through* the conn |
| 12:51.18 | brlcad | because the conn is rightly const |
| 12:51.26 | d-lo | ah, Ic. |
| 12:52.36 | brlcad | so you have a switch somewhere -- in a class or on the stack or malloced on the heap somewhere, you passed a pointer to it to libpkg, which then set it in the conn during dispatch |
| 12:53.01 | brlcad | so you can still modify the switch, just not through the pointer that was given to libpkg |
| 12:53.37 | brlcad | the pkg mod I have in mind get around that problem by making it a callable parameter |
| 12:54.04 | brlcad | but that will be a slightly more invasive api change (for the better) |
| 13:29.33 | *** join/#brlcad akafubu (~akafubu@unaffiliated/akafubu) | |
| 13:30.50 | *** join/#brlcad akafubu (~akafubu@unaffiliated/akafubu) | |
| 14:34.04 | starseeker | kicks self as he spots the PATH_SUFFIXES option to FIND_* in CMake |
| 14:34.06 | starseeker | auuuuugh |
| 14:34.23 | starseeker | that would have saved a few lumps |
| 14:50.40 | starseeker | oh well, FindTCL needed cleanup anyhow |
| 14:56.48 | CIA-2 | BRL-CAD: 03davidloman * r40733 10/rt^3/trunk/ (include/PkgClient.h src/libPkgCpp/PkgClient.cxx): |
| 14:56.48 | CIA-2 | BRL-CAD: Add in a getCallbackTable for verifying the contents of the passed in |
| 14:56.48 | CIA-2 | BRL-CAD: pkg_switch* table. Dropped setCallbackTable since the callback table is being |
| 14:56.48 | CIA-2 | BRL-CAD: set prior to PkgClient creation. Exposed pkg_flush as a PkgClient method. |
| 14:57.54 | CIA-2 | BRL-CAD: 03davidloman * r40734 10/rt^3/trunk/ (include/PkgServer.h src/libPkgCpp/PkgServer.cxx): Added in callback table as an arg to connectToHost and waitForClient |
| 15:01.48 | CIA-2 | BRL-CAD: 03davidloman * r40735 10/rt^3/trunk/ (include/Portal.h src/libNet/Portal.cxx): |
| 15:01.49 | CIA-2 | BRL-CAD: Exposed PkgClient::flush() with Portal::flush() in an attempt to get the writes |
| 15:01.49 | CIA-2 | BRL-CAD: working. Added a handshake logic checker in to Portal::handleNetMsg() to |
| 15:01.49 | CIA-2 | BRL-CAD: prevent two portals from continually sending RemoteNodenameSetMsgs to eachother. |
| 15:01.49 | CIA-2 | BRL-CAD: Removed the pkg_switch table generation from Portal cstr since it is generated |
| 15:01.49 | CIA-2 | BRL-CAD: prior to Portal init. Removed some debug statements and added yet others to |
| 15:01.50 | CIA-2 | BRL-CAD: support continuing t-shooting. |
| 15:03.07 | CIA-2 | BRL-CAD: 03davidloman * r40736 10/rt^3/trunk/src/libNet/PortalManager.cxx: Complete revamp of selector loop. Better logic and flow now. Added in pkg_switch table generation to both incoming and outgoing connection logic. |
| 15:04.08 | CIA-2 | BRL-CAD: 03davidloman * r40737 10/rt^3/trunk/tests/libpkgcpp/pkgcppTest.cxx: Change libpkgcpp test to reflect change in pkg_switch table generation logic. |
| 15:05.01 | CIA-2 | BRL-CAD: 03davidloman * r40738 10/rt^3/trunk/tests/libNet/libNetTest.cxx: Added logging statements in libNet test. |
| 15:12.00 | CIA-2 | BRL-CAD: 03davidloman * r40739 10/rt^3/trunk/src/libNet/Portal.cxx: Missed a file for r40735 |
| 16:06.41 | CIA-2 | BRL-CAD: 03starseeker * r40740 10/rt^3/trunk/src/other/subversion/ (1204 files in 135 dirs): |
| 16:06.42 | CIA-2 | BRL-CAD: Don't wire it in yet, but get the subversion code and its apr requirements into |
| 16:06.42 | CIA-2 | BRL-CAD: the main rt3 module. At the same time, start working on a FindAPR module to |
| 16:06.42 | CIA-2 | BRL-CAD: conditionalize the eventual ExternalProject logic for apr. Need to check the |
| 16:06.42 | CIA-2 | BRL-CAD: FindSUBVERSION module to see if it spots subversion's libraries. |
| 16:20.30 | *** join/#brlcad mafm_ (~mafm@83.49.86.17) | |
| 16:54.26 | CIA-2 | BRL-CAD: 03davidloman * r40741 10/rt^3/trunk/ (3 files in 2 dirs): |
| 16:54.27 | CIA-2 | BRL-CAD: Introduce PortalProxy. pkg needs to have its pkg_switch table filled out |
| 16:54.27 | CIA-2 | BRL-CAD: entirely prior to the creation of a pkg_conn, and current architecture has a |
| 16:54.27 | CIA-2 | BRL-CAD: need for setting parts of the pkg_switch table after pkg_conn creation. |
| 16:54.27 | CIA-2 | BRL-CAD: PortalProxy provides a simple way around this chicken/egg issue. |
| 17:40.03 | CIA-2 | BRL-CAD: 03davidloman * r40742 10/rt^3/trunk/ (8 files in 4 dirs): Fixed the issue with pkg_switch not retaining values. Variable scope ftw. |
| 17:42.54 | CIA-2 | BRL-CAD: 03brlcad * r40743 10/brlcad/trunk/src/librt/primitives/bot/bot.c: quell verbose linux warnings. init bin values to 0, removed unused vars. |
| 17:43.26 | CIA-2 | BRL-CAD: 03davidloman * r40744 10/rt^3/trunk/ (7 files in 4 dirs): WS, Formatting. |
| 17:58.53 | CIA-2 | BRL-CAD: 03brlcad * r40745 10/brlcad/trunk/src/fbserv/server.c: ws indent style cleanup. add new fourth parameter to the pkg switch table, setting user_data to NULL. |
| 18:11.37 | CIA-2 | BRL-CAD: 03davidloman * r40746 10/rt^3/trunk/src/libNet/ (NetMsgFactory.cxx Portal.cxx PortalManager.cxx): Portal<->Portal handshaking is working now. |
| 18:13.31 | CIA-2 | BRL-CAD: 03brlcad * r40747 10/brlcad/trunk/src/fbserv/server.c: quell size_t warnings, cleanup formward function decls, and check pcp for nullity. |
| 18:15.04 | CIA-2 | BRL-CAD: 03davidloman * r40748 10/rt^3/trunk/src/libNet/PortalManager.cxx: Remove writefds from selector loop completely. |
| 18:21.58 | CIA-2 | BRL-CAD: 03davidloman * r40749 10/rt^3/trunk/ (include/NetMsgTypes.h src/libNet/NetMsgFactory.cxx): Add KEEPALIVE msg type. |
| 18:30.25 | CIA-2 | BRL-CAD: 03davidloman * r40750 10/rt^3/trunk/ (include/ControlledThread.h src/utility/ControlledThread.cxx): Made ControlledThread::shutdown() take an optional blocking boolean. Default is true. Shutdown() now blocks until thread has terminated. |
| 18:31.22 | CIA-2 | BRL-CAD: 03davidloman * r40751 10/rt^3/trunk/src/libNet/Portal.cxx: WS, Formatting. |
| 18:32.42 | CIA-2 | BRL-CAD: 03davidloman * r40752 10/rt^3/trunk/include/PortalManager.h: Clay: WS, Formatting. |
| 18:34.24 | CIA-2 | BRL-CAD: 03davidloman * r40753 10/rt^3/trunk/src/libNet/PortalManager.cxx: quick typo fix. Shoulda been INFO not ERROR! |
| 18:34.59 | CIA-2 | BRL-CAD: 03brlcad * r40754 10/brlcad/trunk/configure.ac: more warnings that would be useful and interesting to enable (maintenance task for later) |
| 18:38.48 | CIA-2 | BRL-CAD: 03davidloman * r40755 10/rt^3/trunk/ (4 files in 3 dirs): Introduce RUALIVE and IMALIVE message types. Make Portal respond to RUALIVE with an IMALIVE msg. |
| 18:40.39 | CIA-2 | BRL-CAD: 03davidloman * r40756 10/rt^3/trunk/src/libNet/Portal.cxx: Forgot an include! |
| 18:51.21 | CIA-2 | BRL-CAD: 03davidloman * r40757 10/rt^3/trunk/src/libNet/Portal.cxx: Forgot an include! |
| 18:52.23 | CIA-2 | BRL-CAD: 03davidloman * r40758 10/rt^3/trunk/src/libNet/PortalManager.cxx: Forgot to init fdmax. was causing some trippy errors. Also, added logpoint to note when PortalManager enters/exits its run loop |
| 18:54.18 | CIA-2 | BRL-CAD: 03davidloman * r40759 10/rt^3/trunk/src/libNet/PortalManager.cxx: Comment out some log points. |
| 18:54.41 | starseeker | seems to be for real - the OpenOffice.org community is striking out on their own |
| 18:59.49 | CIA-2 | BRL-CAD: 03davidloman * r40760 10/rt^3/trunk/src/libNet/Portal.cxx: Comment out some log points. |
| 19:00.48 | CIA-2 | BRL-CAD: 03davidloman * r40761 10/rt^3/trunk/src/libNet/PortalManager.cxx: Made the PortalManager remove the FD mapping when a connection drops. Comment out some log points. |
| 19:02.27 | CIA-2 | BRL-CAD: 03davidloman * r40762 10/rt^3/trunk/ (3 files in 2 dirs): Drop PortalProxy. Used a mo betta solution. Thanks Sean! |
| 19:09.58 | ``Erik | huh, found a wad of simh stuff, I think one of these is my vgr image |
| 19:10.21 | starseeker | swwweeet |
| 19:11.05 | ``Erik | and I need new tires :/ |
| 19:11.21 | starseeker | ow |
| 19:11.58 | ``Erik | (would explain this mornings surprise sideways driving, though) |
| 19:12.17 | d-lo | that needs explaining? |
| 19:12.21 | d-lo | =D |
| 19:14.03 | ``Erik | well, the surprise part, yeah |
| 19:14.13 | ``Erik | bump in a turn got my tail a bit loose :) |
| 19:14.53 | d-lo | contemplates a 'loose tail' joke.... |
| 19:15.03 | ``Erik | that's why I'm sitting on a donut, yo |
| 19:16.27 | CIA-2 | BRL-CAD: 03davidloman * r40763 10/rt^3/trunk/ (4 files in 2 dirs): Stub in NetMsgRouter. It shall be used for.... routing NetMsgs... |
| 19:18.52 | *** join/#brlcad mafm (~mafm@83.49.86.17) | |
| 19:27.45 | CIA-2 | BRL-CAD: 03brlcad * r40764 10/brlcad/trunk/src/libfb/ (fbserv_obj.c if_remote.c): pkg_switches now take a fourth parameter. set to NULL. |
| 19:46.53 | CIA-2 | BRL-CAD: 03davidloman * r40765 10/rt^3/trunk/ (include/NetMsgRouter.h src/libNet/NetMsgRouter.cxx): Implement registration of NetMsg types with respective NetMsgHandlers. Implement routing of NetMsgs |
| 19:48.48 | CIA-2 | BRL-CAD: 03davidloman * r40766 10/rt^3/trunk/src/libNet/Portal.cxx: Link the Portal::Springboard to NetMsgRouter::routeMsg() |
| 19:51.54 | CIA-2 | BRL-CAD: 03davidloman * r40767 10/rt^3/trunk/ (include/NetMsgRouter.h src/libNet/NetMsgRouter.cxx): Stub in a hook for a method that will auto-register type/handler combos upon first access of NetMsgRouter |
| 19:54.36 | CIA-2 | BRL-CAD: 03davidloman * r40768 10/rt^3/trunk/include/NetMsg.h: Fix QT includes to have full path |
| 19:55.50 | *** join/#brlcad Zaebos (~irc@pd95b7f5e.dip0.t-ipconnect.de) | |
| 19:57.05 | CIA-2 | BRL-CAD: 03davidloman * r40769 10/rt^3/trunk/ (include/NetMsg.h src/libNet/netMsg/NetMsg.cxx): Add a getter for NetMsg::origin |
| 20:03.08 | CIA-2 | BRL-CAD: 03davidloman * r40770 10/rt^3/trunk/src/libNet/NetMsgRouter.cxx: Add in debug printing. |
| 20:17.07 | CIA-2 | BRL-CAD: 03brlcad * r40771 10/brlcad/trunk/misc/win32-msvc8/Makefile.am: terrain was renamed to wavy |
| 20:28.01 | *** join/#brlcad R0b0t1 (~Enigma@64-136-219-55.dyn.everestkc.net) | |
| 20:28.01 | *** join/#brlcad R0b0t1 (~Enigma@unaffiliated/r0b0t1) | |
| 20:46.01 | CIA-2 | BRL-CAD: 03starseeker * r40772 10/rt^3/trunk/cmake/FindSVNLIBS.cmake: Gonna probably need a find routine for SVN libraries - I suppose we really shouldn't be building them if they're present |