| 00:48.55 | *** join/#brlcad juanman (~quassel@unaffiliated/juanman) | |
| 00:55.58 | brlcad | starseeker: shouldn't need to make any accommodations in the build system for LFS specifically |
| 00:56.11 | brlcad | that was the point and why autotools build work was paused too |
| 00:56.43 | brlcad | the point was to unroll back to a good ol' fashioned /usr/brlcad install hierarchy controlled by a prefix destination |
| 00:57.39 | brlcad | there would be some completely secondary tool/script that would create all of the symbolic versioned infrastructure |
| 01:00.48 | *** join/#brlcad juan_man (~quassel@unaffiliated/juanman) | |
| 01:01.00 | brlcad | rather, the /usr/brlcad/rel-* /usr/brlcad/dev-* that we have now would stay, the share/brlcad/VERSION would go back to just share, and I think the rest stays the same |
| 01:01.41 | brlcad | so it's not all for naught, it just simplifies greatly |
| 01:02.06 | brlcad | in fact, you could probably follow http://www.gnu.org/prep/standards/html_node/Directory-Variables.html for defaults if you want to add variable toggle parallels |
| 01:02.49 | brlcad | dealing with LFS then becomes a simple matter of writing that script to create symbolic links -- thinking about it I think the brlcad-config script is probably a perfect place to put it even |
| 01:04.14 | CIA-23 | BRL-CAD: 03brlcad * r51550 10/brlcad/trunk/misc/brlcad-config.in: eliminate authorship from file |
| 01:06.03 | brlcad | hm, we may need to rewrite this somewhat for cmake |
| 01:11.25 | starseeker | to be honest, I hadn't messed with brlcad-config much - no idea if it's working |
| 01:11.58 | starseeker | is in the middle of another change - after that, will strip out the LFS stuff |
| 01:13.09 | brlcad | oh I'm almost sure it's not working, it's an autoconf template |
| 01:13.16 | starseeker | heh - oops |
| 01:14.32 | brlcad | so curious, is MAN_DIR a configurable variable or just a var used internally? |
| 01:15.24 | starseeker | configurable |
| 01:15.25 | brlcad | I could see them as advanced options with familiar GNU defaults, but actualy wasn't expecting it |
| 01:15.39 | brlcad | oh, that's neat then |
| 01:15.52 | starseeker | after I finally got done with the surgery, you could set MAN_DIR=man and have all the man pages in a toplevel man directory |
| 01:16.11 | brlcad | what about MAN_DIR=/path/to/man |
| 01:16.12 | starseeker | only drawback is I can't pass in CMake variables on the command line (e.g MAN_DIR=${DATA_DIR}/man won't work) |
| 01:16.30 | starseeker | um. haven't tested absolute path - probably a bad idea |
| 01:16.35 | brlcad | big whoop, can't do that with autoconf (easily) either |
| 01:16.51 | brlcad | autoconf is expected to be full paths |
| 01:16.53 | starseeker | in fact, VERY bad idea - no way to distinguish between build dir and install dir |
| 01:17.34 | starseeker | yeah, CMAKE_INSTALL_PATH and CMAKE_BINARY_DIR are where the abosolute stuff lives for output - rest is relative to that |
| 01:20.00 | CIA-23 | BRL-CAD: 03starseeker * r51551 10/brlcad/trunk/ (13 files in 12 dirs): Take the DATA_DIR assumption out of DOC_DIR targets - DOC_DIR output still defaults to using DATA_DIR, but is no longer forced to. |
| 01:20.02 | starseeker | s/CMAKE_INSTALL_PATH/CMAKE_INSTALL_PREFIX |
| 01:21.08 | brlcad | so lets see -- for the ones we care about -- by default gnu does: prefix=/usr bindir=prefix/bin datadir=prefix/share sysconfdir=prefix/etc includedir=prefix/include libdir=prefix/lib mandir=datadir/man and docdir=datadir/doc |
| 01:27.36 | starseeker | I'm not actually using sysconfdir for anything right now... probably something getting stuck in data that should go there? |
| 01:28.20 | CIA-23 | BRL-CAD: 03starseeker * r51552 10/brlcad/trunk/ (CMakeLists.txt src/other/CMakeLists.txt): Strip out attempt at LFS specific prefix settings |
| 01:29.18 | starseeker | ah ha |
| 01:34.41 | CIA-23 | BRL-CAD: 03starseeker * r51553 10/brlcad/trunk/src/other/libtermlib/CMakeLists.txt: termap goes in CONF_DIR, not DATA_DIR/etc |
| 01:34.48 | starseeker | there we go |
| 01:36.10 | CIA-23 | BRL-CAD: 03starseeker * r51554 10/brlcad/trunk/CMakeLists.txt: Stray variable left over from early days. |
| 01:37.41 | starseeker | I think we now match those dirs - the issue remains though, that there isn't a handy version consistency check for bu_brlcad_data |
| 01:39.05 | starseeker | when DATA_DIR incorporates the version, stuff breaks when you try to use the wrong data dir - it would be nice if bu_brlcad_data could verify the version with some standard file included in DATA_DIR/BRL-CAD.version or some such |
| 01:53.00 | brlcad | starseeker: what do you mean "try to use the wrong data dir"? |
| 02:02.25 | starseeker | let's say someone installs 7.20.2 into /usr/brlcad and 7.22.0 into /usr/brlcad/rel-7.22.0 - wouldn't that result in 7.22.0 trying to use 7.20.2's data? |
| 02:03.06 | starseeker | or, worse, BRLCAD_DATA environment variable gets messed with without the user being consciously aware of it |
| 02:07.30 | CIA-23 | BRL-CAD: 03starseeker * r51555 10/brlcad/trunk/src/other/step/src/express/CMakeLists.txt: hmm - exppp tried to build before fedex, and didn't have version_string. Make the version string dependency trigger on expres build |
| 02:21.48 | brlcad | that should not result in 7.22.0 using 7.20.2's data or vice versa |
| 02:25.28 | starseeker | makes a note to do some tests later... but cool, as long as we don't get weird obscure errors from incorrect tcl scripts getting loaded I'm good |
| 02:25.53 | brlcad | if it does, some bug got injected because that was all tested at one point |
| 02:26.24 | starseeker | sounds good |
| 02:41.34 | brlcad | for data, it should basically it should be trying 1) env var override, 2) compile-time datadir, 3) root-relative datadir, 4) rootdir, and last and least, 5) current dir |
| 02:41.51 | brlcad | so unless environment var is set, it should always find the compile-time path |
| 02:42.14 | brlcad | 3, 4, and 5 are in place to support relocation |
| 02:52.55 | brlcad | if you move brl-cad, it should basically default to root-relative datadir, which will be a similar 1) env var (no), 2) compile-time (no), 3) run-time (yes) .. before even getting to 4) /usr/brlcad and finally 5) current dir |
| 03:20.58 | starseeker | but in principle, an env var override could feed it a bad data dir? I suppose that's inevitable and maybe even intended, but it would be silent and attempt to run with share and fail to find anything with share/brlcad/VERSION |
| 03:21.50 | starseeker | I suppose though, the principle is if you're mucking with env variables you know what you're doing |
| 03:21.59 | starseeker | winces - in theory anyway |
| 03:25.39 | starseeker | makes a note to add sanity checking to BIN_DIR and friends - full paths suppled to them should actually be an error |
| 03:35.03 | CIA-23 | BRL-CAD: 03starseeker * r51556 10/brlcad/trunk/CMakeLists.txt: reject full paths for BIN_DIR, LIB_DIR and friends. |
| 03:38.51 | CIA-23 | BRL-CAD: 03starseeker * r51557 10/brlcad/trunk/misc/CMake/BRLCAD_Summary.cmake: Show LIB_DIR as well. |
| 03:40.14 | brlcad | starseeker: yeah, that's by design -- if BRLCAD_ROOT or BRLCAD_DATA are set, that is the path to use by definition, always |
| 03:43.52 | brlcad | it required active user participation to specify |
| 03:44.34 | brlcad | at best we either don't allow an override or we attempt to second guess something the user told us to do (VERY BAD) |
| 03:45.20 | starseeker | is going to start toting a nurf bat for scripting lessons... |
| 03:46.31 | brlcad | users really shouldn't ever be setting either, shouldn't even know to try and set it any more unless we tell them to |
| 03:46.55 | starseeker | apparently, old scripts never die |
| 03:47.46 | starseeker | write once, support forever |
| 03:54.39 | brlcad | so that is a good point |
| 03:55.18 | brlcad | we could put some code in place to detect a mismatch and at least print a message |
| 04:04.29 | *** join/#brlcad louipc (~louipc@archlinux/fellow/louipc) | |
| 04:21.35 | andrei | brlcad : when you have time could you please look into red black tree unit test? It s still pending and it should require minimal feedback/correction. http://sourceforge.net/tracker/index.php?func=detail&aid=3510515&group_id=105292&atid=640804 |
| 04:28.59 | andrei | oh and by the way : I found out that the script output was in csv format ( well, sort of ) |
| 04:29.29 | andrei | Instead of having (package_size, file_size, elapsed_time ) I have each on one line, I need to figure out a way to collect them |
| 05:12.11 | brlcad | excellent, thanks for the reminder (you might have to remind a couple more times to keep the priority up) |
| 05:41.09 | CIA-23 | BRL-CAD: 03phoenixyjll * r51558 10/brlcad/trunk/src/librt/ (opennurbs_ext.cpp opennurbs_ext.h): Fit the points into NURBS curves using polyline approximation. (WIP) |
| 05:44.52 | CIA-23 | BRL-CAD: 03Phoenix 07http://brlcad.org * r4184 10/wiki/User:Phoenix/GSoc2012/Reports: /* Week 9 */ |
| 07:08.38 | *** join/#brlcad andrei (andrei@188.25.160.102) | |
| 08:23.01 | *** join/#brlcad d_rossberg (~rossberg@BZ.BZFLAG.BZ) | |
| 10:17.37 | *** join/#brlcad andrei (andrei@188.25.160.67) | |
| 10:41.05 | *** join/#brlcad Stattrav (u3131@gateway/web/irccloud.com/x-jpqzeciblydvioqx) | |
| 11:39.40 | CIA-23 | BRL-CAD: 03n_reed * r51559 10/brlcad/trunk/src/librt/primitives/sketch/sketch_tess.cpp: change label name to appease MSVC |
| 12:02.55 | starseeker | brlcad: that was my thought - do what they say, but warn if we're trying to use a data dir from another version (doesn't have to be loud and garish even - just something so we can potentially know what's happening before the obscure bugs appear) |
| 12:28.28 | CIA-23 | BRL-CAD: 03n_reed * r51560 10/brlcad/trunk/src/other/clipper/clipper.hpp: need to export non-member functions |
| 12:40.45 | ``Erik | andrei: if you want to collapse lines, 'xargs' is one approach. default command is echo, so maybe something like: (cmd 1; cmd 2; cmd 3) | xargs >> file |
| 12:52.53 | CIA-23 | BRL-CAD: 03n_reed * r51561 10/brlcad/trunk/src/other/perplex/perplex_template.c: finish incomplete substitution from r51480 |
| 13:00.10 | andrei | ah, I did 3 echo -n instead |
| 13:00.42 | andrei | I ll provide the csv format data today but I still can't find any explanation why isn't any time > 1 second |
| 13:01.01 | andrei | transfering 8mb with 1byte packages takes about 300 sec for example |
| 13:29.08 | ``Erik | -n isn't posix and doesn't show up in bsd echo (there's a printf command to handle it), fwiw |
| 13:30.00 | ``Erik | do you have your scripts in, say, your personal github repo? |
| 13:30.49 | ``Erik | or the tpkg command? |
| 13:34.11 | ``Erik | (is the tpkg you're using the one in the BRL-CAD svn?) |
| 14:24.50 | *** join/#brlcad ksuzee (~ksu@193.151.105.83) | |
| 14:55.45 | CIA-23 | BRL-CAD: 03n_reed * r51562 10/brlcad/trunk/src/other/step/src/cleditor/instmgr.h: remove conflicting EntityClassName declaration |
| 15:37.40 | CIA-23 | BRL-CAD: 03Crdueck 07http://brlcad.org * r4185 10/wiki/User:Crdueck/log: |
| 16:43.37 | CIA-23 | BRL-CAD: 03n_reed * r51563 10/brlcad/trunk/src/other/step/ (9 files in 6 dirs): rename getopt to sc_getopt and include it as part of base lib; SCL git 1a60cf9 and dd93a96 |
| 16:57.23 | *** join/#brlcad Stattrav (u3131@gateway/web/irccloud.com/x-jpqzeciblydvioqx) | |
| 16:57.23 | *** join/#brlcad andrei (andrei@188.25.160.67) | |
| 16:57.23 | *** join/#brlcad Mahi (~Mahi@ec2-50-17-113-35.compute-1.amazonaws.com) | |
| 17:06.12 | ``Erik | http://www.youtube.com/watch?v=yO7MWuJ7zLA |
| 17:42.18 | CIA-23 | BRL-CAD: 03crdueck * r51564 10/brlcad/trunk/src/librt/primitives/sketch/sketch_tess.cpp: use signed curvature to find inflection point in bezier_inflection(), change test condition for approx_bezier(), fix some errors |
| 17:50.11 | *** join/#brlcad crdueck (~cdk@d173-238-127-19.home4.cgocable.net) | |
| 18:58.46 | CIA-23 | BRL-CAD: 03n_reed * r51565 10/brlcad/trunk/src/librt/primitives/sketch/sketch_tess.cpp: s/loop/bez_to_carcs_loop/ to match other renamed labels |
| 19:25.05 | CIA-23 | BRL-CAD: 03n_reed * r51566 10/brlcad/trunk/src/other/step/ (9 files in 9 dirs): update dll export macros and remove local definitions; SCL git 7be54f3 |
| 19:28.55 | *** join/#brlcad crdueck (~cdk@d173-238-127-19.home4.cgocable.net) | |
| 19:57.39 | *** join/#brlcad sharvey_ (~sharvey@xi.metaether.net) | |
| 19:58.35 | *** join/#brlcad yiyus_ (1242712427@je.je.je) | |
| 20:17.53 | *** join/#brlcad crdueck (~cdk@d173-238-127-19.home4.cgocable.net) | |
| 20:51.58 | CIA-23 | BRL-CAD: 03starseeker * r51567 10/brlcad/trunk/ (4 files in 4 dirs): Tweak handling of perplex template |
| 21:03.39 | CIA-23 | BRL-CAD: 03starseeker * r51568 10/brlcad/trunk/src/other/step/cmake/SCL_Utils.cmake: fix SCL cmake logic |
| 21:10.33 | CIA-23 | BRL-CAD: 03starseeker * r51569 10/brlcad/trunk/src/libpkg/tpkg.c: this style of quellage doesn't work with clang - need UNUSED |
| 21:13.56 | starseeker | ``Erik: you like bit stuff - any idea what this is about? http://pastebin.mozilla.org/1706681 |
| 21:55.20 | ``Erik | saw that before, think it's the compiler being goofy (only happens on a certain platform/flag combination or something), I think I fixed it by breaking the a &= b into a = a & b or something |
| 21:55.48 | ``Erik | by 'fixed', quelled the warning, should be no change in behavior unless the compiler does something wonky |
| 23:07.19 | *** join/#brlcad dtidrow (~dtidrow@c-68-84-167-135.hsd1.mi.comcast.net) | |
| 23:38.46 | CIA-23 | BRL-CAD: 03Cprecup 07http://brlcad.org * r4186 10/wiki/User:Cprecup/GSoC2012_progress: 16-17/07/2012 - Switching from hash tables to lists of names. Introduced a master "objects" hash table. |