| 00:24.17 | *** join/#brlcad starseek1r (~starseeke@66-118-151-70.static.sagonet.net) |
| 00:27.58 | *** join/#brlcad mpictor_ (~mark@c-67-177-102-131.hsd1.in.comcast.net) |
| 00:29.42 | *** join/#brlcad Ch3ck_ (~Ch3ck@66-118-151-70.static.sagonet.net) |
| 00:43.23 | *** join/#brlcad ejno (~ejno@66-118-151-70.static.sagonet.net) |
| 00:47.27 | *** join/#brlcad ``Erik (~erik@pool-74-103-94-19.bltmmd.fios.verizon.net) |
| 00:50.30 | *** join/#brlcad kanzure (~kanzure@131.252.130.248) |
| 00:55.41 | *** join/#brlcad KimK (~Kim__@ip24-255-223-153.ks.ks.cox.net) |
| 00:58.24 | *** join/#brlcad hickoryknoll (~hickorykn@66-118-151-70.static.sagonet.net) |
| 00:58.50 | *** join/#brlcad ejno (~ejno@66-118-151-70.static.sagonet.net) |
| 01:00.48 | *** join/#brlcad witness___ (uid10044@gateway/web/irccloud.com/x-knjvtijecggsarfb) |
| 01:01.01 | *** join/#brlcad yiyus (1242712427@je.je.je) |
| 18:13.01 | *** join/#brlcad infobot (~infobot@rikers.org) |
| 18:13.01 | *** topic/#brlcad is BRL-CAD || http://brlcad.org || logs: http://ibot.rikers.org/%23brlcad/ || GCI winners: Jacob Burroughs and Peter Amidon! Thanks to all of our participants for the awesome work; stay in touch! || We're applying to GSoC 2014 |
| 18:21.47 | *** join/#brlcad chick (~chick@195.24.220.16) |
| 18:33.56 | brlcad | ls -la |
| 18:58.20 | brlcad | ``Erik: no notify love? |
| 18:58.29 | ``Erik | just started looking into it |
| 18:58.35 | brlcad | ah, cool |
| 19:01.20 | *** join/#brlcad jasleen (75fde8a3@gateway/web/freenode/ip.117.253.232.163) |
| 19:01.35 | *** join/#brlcad Notify (~notify@66-118-151-70.static.sagonet.net) |
| 19:02.23 | ``Erik | 76 messages to push out O.o |
| 19:03.02 | Notify | 03BRL-CAD:n_reed * 59647 brlcad/trunk/src/libbu/bitv.c: fix 'var may be used uninitialized' warning |
| 19:03.13 | Notify | 03BRL-CAD:r_weiss * 59648 (brlcad/trunk/include/bio.h brlcad/trunk/include/bu.h and 2 others): Change macro IGNORE to BU_IGNORE to prevent a conflict with the IGNORE macro in the Windows header winbase.h. |
| 19:03.15 | Notify | 03BRL-CAD:r_weiss * 59649 brlcad/trunk/src/libbu/tests/bu_semaphore.c: Quiet a windows build warning. |
| 19:03.20 | Notify | 03BRL-CAD:r_weiss * 59650 brlcad/trunk/src/libfb/fbserv_obj.c: Quiet a windows build warning. |
| 19:03.24 | Notify | 03BRL-CAD:r_weiss * 59651 (brlcad/trunk/include/config_win.h.in brlcad/trunk/src/libbu/getcwd.c): Windows needs the direct.h header for the _getcwd function. |
| 19:03.29 | Notify | 03BRL-CAD:r_weiss * 59652 brlcad/trunk/include/fbio.h: Windows needs the header io.h in the framebuffer library. |
| 19:03.32 | Notify | 03BRL-CAD:r_weiss * 59653 brlcad/trunk/include/config_win.h.in: Windows needs the float.h header for the _isnan and _finite functions. |
| 19:03.34 | Notify | 03BRL-CAD:brlcad * 59654 brlcad/trunk/include/fbio.h: there is no apparent io.h symbol in fbio.h, please describe the need (and bio.h should be used instead of io.h) |
| 19:03.39 | Notify | 03BRL-CAD:r_weiss * 59655 (brlcad/trunk/include/bu.h brlcad/trunk/src/libbu/vls.c): It appears vls_offset should be a signed size_t. This change also quiets a windows build warning. |
| 19:03.45 | Notify | 03BRL-CAD:brlcad * 59656 (brlcad/trunk/misc/CMake/BRLCAD_Targets.cmake brlcad/trunk/misc/CMake/BRLCAD_Util.cmake): ideally, including one of these resource files should not cause tests to be performed. move the test for Wno-error and symlink logic into the respective functions where they're actually used. caching should still mean they're only run once and they're localized where needed. |
| 19:05.48 | *** join/#brlcad ChanServ (ChanServ@services.) |
| 19:05.48 | *** mode/#brlcad [+o ChanServ] by dickson.freenode.net |
| 19:14.01 | brlcad | ``Erik: woot |
| 19:14.18 | brlcad | get stuck because of the dns change? |
| 19:14.35 | *** join/#brlcad jasleen (~chatzilla@117.253.232.163) |
| 19:15.10 | brlcad | ``Erik: up for gsoc again? ... this is our 10-year open source anniversary, going to be a big one :) |
| 19:15.17 | brlcad | hi jasleen |
| 19:15.36 | jasleen | hello |
| 19:15.58 | ``Erik | d'no, suspect it was related to the ddos against freenode last night, mebbe they did a shutdown/relink on the server it was connected to and the disconnect didn't seem abnormal enough to trigger the reconnect, the 'overmind' thread was marked as having exited normally |
| 19:16.42 | ``Erik | sure, I'll do gsoc (and this time, I'll accept the stipend, w00t!) |
| 19:17.26 | ``Erik | do you need a bio blurb, or is the old one ok? |
| 19:23.06 | brlcad | old one's okay, just need to know your username |
| 19:27.12 | ``Erik | created, erikg |
| 19:27.57 | ``Erik | (am I secondary admin, too?) |
| 19:30.21 | *** join/#brlcad jasleen (~chatzilla@117.253.232.163) |
| 19:37.36 | *** join/#brlcad javampire (~Csaba@p4FF7143C.dip0.t-ipconnect.de) |
| 19:58.18 | *** join/#brlcad __javampire__ (~Csaba@p4FF7143C.dip0.t-ipconnect.de) |
| 19:58.38 | *** part/#brlcad __javampire__ (~Csaba@p4FF7143C.dip0.t-ipconnect.de) |
| 20:14.09 | *** join/#brlcad Notify (~notify@66-118-151-70.static.sagonet.net) |
| 20:14.53 | Notify | 03BRL-CAD:r_weiss * 59657 brlcad/trunk/src/libged/joint.c: Missed this change in r59655 when changing vls_offset to ssize_t. |
| 20:16.27 | Notify | 03BRL-CAD:r_weiss * 59659 brlcad/trunk/src/libbu/tests/test_funcs.c: Removed the include of stdbool.h. Not used and breaks windows build. |
| 20:16.58 | Notify | 03BRL-CAD:r_weiss * 59658 (brlcad/trunk/CMakeLists.txt brlcad/trunk/src/libbu/getcwd.c): Changed logic for testing for and using the windows header direct.h. |
| 20:17.00 | Notify | 03BRL-CAD:brlcad * 59660 (brlcad/trunk/src/libfb/fb_generic.c brlcad/trunk/src/libfb/fbserv_obj.c brlcad/trunk/src/libfb/if_tk.c): need to include bio.h for standard I/O functions. coincidentally simplifies the files because of unistd and stdio inclusions getting consolidated. |
| 20:37.33 | *** join/#brlcad javampire (~Csaba@p4FF7143C.dip0.t-ipconnect.de) |
| 20:38.36 | javampire | kanzure: let me know if you're around (I'm ncsaba, but that nick is registered so I will use javampire from now on) |
| 20:40.16 | javampire | I got to the conclusion that python-brlcad needs to include quite a few macros to work properly, especially the constants definitions |
| 20:40.59 | javampire | the default is to skip them, and that is in fact OK - but we need to be able to give ctypesgen a list of macros we want to have |
| 20:41.18 | javampire | I checked the ctypesgen code and I think it's relatively easy to hack that in |
| 20:42.09 | javampire | but before I can go ahead and do that, I want to have the config file version merged in to the master branch - it is a LOT easier to handle new configuration that way |
| 20:42.49 | javampire | kanzure: so my question is - if I figure out the windows problems, is that enough to go ahead and merge the config file code ? |
| 20:43.45 | javampire | I'm writing this from the windows VM, and will go ahead and debug the problems, hopefully getting to a working solution |
| 20:44.45 | kanzure | i am always aronud =) |
| 20:44.48 | kanzure | also around |
| 20:44.52 | javampire | :-) |
| 20:45.05 | kanzure | the macros get hard to parse |
| 20:45.14 | kanzure | ctypesgencore could have an include/exclude list for macros maybe |
| 20:45.15 | javampire | well I don't want them all |
| 20:45.41 | kanzure | to be honest i would rather just force brlcad to not use macros for core library functions |
| 20:45.50 | javampire | but for example the magic values for the primitive types internal representation, I would really like to have them directly from the brl-cad code and not copy-paste |
| 20:46.01 | kanzure | have you been able to test the config file on both linux and windows? |
| 20:46.21 | javampire | sure, it works fine but there are other problems |
| 20:46.55 | javampire | on windows only libbu, libbn and brep work, rt fails |
| 20:47.24 | javampire | that's happening with the original code too (no config file version), but I still want to debug it |
| 20:47.41 | kanzure | but iirc rt is not being included |
| 20:48.09 | javampire | it is on linux, and it works to |
| 20:48.10 | javampire | too |
| 20:48.20 | javampire | on windows I get this: Ctypes does not support the type "long long". Typedef "_off64_t" will not be output |
| 20:48.40 | javampire | and then: _off64_t = long long # /usr/include/sys/_types.h: invalid syntax |
| 20:48.51 | javampire | so it should be excluded, but for some reason it is still output |
| 20:49.14 | javampire | will need to debug it to see what happens |
| 20:49.40 | javampire | BTW, I think that's some cygwin/msys specific thing, as it happens on both |
| 20:50.08 | kanzure | anyway yes i'll merge it |
| 20:50.36 | javampire | well, you can wait till I debug the windows problem, I really want that out the way |
| 20:51.13 | javampire | and I will need to merge the "win-port" branch, only that one works currently on windows |
| 20:51.26 | kanzure | sorry i am not quite following the state |
| 20:51.27 | javampire | it has some escaping of spaces and such things |
| 20:51.32 | kanzure | the current state of the repo is that it does or does not work? |
| 20:51.45 | Notify | 03BRL-CAD:starseeker * 59661 (brlcad/trunk/src/bwish/input.c brlcad/trunk/src/conv/dem-g.c and 40 others): Convert most of the %V vls printf uses in the source to %s and bu_vls_addr. |
| 20:51.46 | kanzure | and what about the state of this branch? https://github.com/kanzure/python-brlcad/pull/12 windows yes/no linux yes/no |
| 20:51.51 | javampire | the master branch as it is will not work on my windows installation |
| 20:52.34 | javampire | the win-port branch works on my windows for bu, bn, brep |
| 20:53.24 | javampire | I will make it work for the rest too, and then merge it to the master |
| 20:54.38 | javampire | in any case I actually have a pretty nasty windows installation, with spaces in the paths, so at least that is tested too |
| 20:55.15 | kanzure | i don't enjoy how the only way to test this is to actually install the python package (this is very broken) |
| 20:55.44 | kanzure | there is a way to bootstrap this to allow non-pip/easyinstall installation |
| 20:55.53 | kanzure | but i haven't implemented it yet |
| 20:56.33 | javampire | I have no expereince with any of this... |
| 20:57.03 | javampire | before installing python-brlcad I never heard of pip |
| 20:59.23 | javampire | OK, then I'll do it like this: fix the windows port, merge it to the master branch, create a new pull request |
| 20:59.37 | javampire | the old one is probably to be dropped |
| 21:00.12 | javampire | I'm also new to git-hub (and git), so I'm also experimenting with all that too |
| 21:01.07 | kanzure | setuptools/pip is just so that users can get bindings generated when they install the package |
| 21:01.13 | kanzure | rather than distributing pre-made bindings with the package |
| 21:01.38 | kanzure | but for testing the binding generation functions themselves, there's really no fundamental reason that the python packaging has to be repeated every attempt |
| 21:06.17 | javampire | well the most time consuming step is the binding generation anyway, so I just re-install the package for each test |
| 21:08.18 | kanzure | reinstallation requires some of the dependencies to be re-downloaded, which is silly for your purposes. i'll work up something soon. |
| 21:09.08 | javampire | well it seems to me it works off-line too, so there should be no downloading... |
| 21:09.27 | javampire | true I do it via the setup.py script and not via pip |
| 21:09.58 | kanzure | ah okay |
| 21:10.44 | javampire | finally I would need to try the pip way too, but I'm not sure how to package, and possibly put somewhere locally for testing ? |
| 21:12.14 | javampire | BTW, pip will also download and then run setup.py, right ? |
| 21:13.10 | javampire | I haven't tested it that way, hopefully it has no problems related to finding the config file |
| 21:13.28 | javampire | how would I package for pip and test it ? |
| 21:14.05 | kanzure | to package: python setup.py dist |
| 21:14.09 | javampire | ok |
| 21:14.24 | kanzure | pip install python-brlcad downloads the last published version of the package from pypi.python.org (this is controlled by me at the moment) |
| 21:14.40 | javampire | is there a way to force it use some local repository ? |
| 21:14.48 | javampire | and some pointers how to set that up ? |
| 21:15.05 | kanzure | devpi-cache or something.. most people don't bother |
| 21:15.19 | kanzure | devpiserver? |
| 21:15.25 | javampire | will look it up |
| 21:16.34 | javampire | ok, it's "-f, --find-links <url>" parameters of pip |
| 21:17.21 | javampire | I'll test it once I debug the cygwin/msys problem |
| 21:17.43 | javampire | BTW, is the current published version working for you on windows ? |
| 21:18.22 | kanzure | i haven't tested in a month =) |
| 21:18.29 | kanzure | we need more automatic tests |
| 21:18.30 | javampire | ok :-) |
| 21:18.56 | javampire | well the windows setup is quite hairy anyway |
| 21:19.14 | javampire | need to install cygwin/msys, and neither is an easy install |
| 21:20.27 | javampire | also there are a few options of python, and the combinations with cygwin and msys, so not sure how easy to test all that |
| 21:21.17 | javampire | currently I have the python coming with cygwin |
| 21:21.53 | javampire | that's at least a somewhat easier choice, you need to install only one suite and it is integrated |
| 21:22.48 | javampire | the combination of separate python + cygwin/mingw was harder to use |
| 21:28.22 | kanzure | if you don't like installing cygwin then you can pickup a vagrant box with it pre-installed |
| 21:28.31 | kanzure | there's lots of veewee/packer builds that give you a windows vm with cygwin already installed |
| 21:35.09 | *** join/#brlcad hickoryk1oll (~hickorykn@66-118-151-70.static.sagonet.net) |
| 21:36.01 | javampire | well I already have it, the idea is that it is not the easiest to install, so python-brlcad on windows will not be for point-and-click type of users |
| 21:36.33 | javampire | it's true it is not really targeting those users either |
| 21:36.58 | kanzure | if you want a point-and-click thing then i suggest making a separate repo with a gui |
| 21:37.13 | kanzure | if you need an exe then i suggest using py2exe or pyinstaller |
| 21:37.20 | javampire | no, I don't want :-) |
| 21:37.20 | kanzure | also there's a way to generate installer exes using setuptools but i forget how |
| 21:37.37 | javampire | I just doubt the usefultness of the effort of porting to windows... |
| 21:37.56 | kanzure | there's a large number of brlcad/windows users :( |
| 21:37.57 | javampire | but it's close to have it so it's OK |
| 21:38.10 | kanzure | (the fact that it works at all is pretty cool) |
| 21:38.15 | javampire | will they also want to script it via python ? |
| 21:38.20 | javampire | possibly... |
| 21:38.24 | kanzure | instead of tcl? yes. yes they do. |
| 21:38.32 | javampire | OK, fair |
| 21:39.40 | javampire | I wonder why the brl-cad windows port was not based on cygwin or mingw ? |
| 21:40.01 | kanzure | brlcad existed before cygwin |
| 21:40.04 | javampire | at least theoretically would have been the easiest thing to do |
| 21:40.06 | javampire | aha |
| 21:43.53 | *** join/#brlcad archivist (~archivist@host81-149-189-98.in-addr.btopenworld.com) |
| 23:11.29 | javampire | OK, I'm progressing with the debug, it's definitely a ctypesgen bug - it recognizes the "long long" as invalid type but still includes the definition for some reason |
| 23:13.08 | javampire | I guess tomorrow I will get to the root of the problem, I have now the data structures caught in flagranti, need only to find who is doing them wrong |
| 23:23.02 | Notify | 03BRL-CAD:carlmoore * 59662 brlcad/trunk/src/conv/nastran-g.c: Remove newline from the Usage; move 'nastran_file =' because we don't need it if we are exiting program; supply default,h,? |