| 00:06.04 | CIA-43 | BRL-CAD: 03johnranderson * r42102 10/brlcad/trunk/src/libbu/simd.c: Since we are using -Wundef option, we must use defined( __SSE__ ) |
| 00:32.49 | CIA-43 | BRL-CAD: 03brlcad * r42103 10/brlcad/trunk/src/libbn/bntester.c: init vars. potential for function_num and test_case_line_num to be accessed without initialization |
| 00:33.08 | CIA-43 | BRL-CAD: 03brlcad * r42104 10/brlcad/trunk/src/librt/primitives/brep/brep.cpp: trailing comma at end of enumerator list makes gcc unhappy |
| 00:33.59 | brlcad | ``Erik: --enable-warnings (before r42099) had two or three extra warnings that weren't enabled by default |
| 00:45.23 | CIA-43 | BRL-CAD: 03brlcad * r42105 10/brlcad/trunk/src/libbn/bntester.c: (log message trimmed) |
| 00:45.24 | CIA-43 | BRL-CAD: this is a really tricky one. quell warnings about variables getting clobbered |
| 00:45.24 | CIA-43 | BRL-CAD: after a longjmp or vfork by making them all static. we can do this here because |
| 00:45.50 | CIA-43 | BRL-CAD: they're all just main() variables fortunately. bu's jump mechanism uses |
| 00:45.51 | CIA-43 | BRL-CAD: setjump, so an implementation could clobber local non-static non-volatile |
| 00:45.51 | CIA-43 | BRL-CAD: variables. this tricking making them static works or it would have also worked |
| 00:45.51 | CIA-43 | BRL-CAD: to wrap the BU_SETJUMP/BU_UNSETJUMP calls into functions that are passed |
| 00:59.50 | CIA-43 | BRL-CAD: 03brlcad * r42107 10/brlcad/trunk/src/libbu/simd.c: no guarantee that __GNUC__ will be defined either. |
| 00:59.52 | CIA-43 | BRL-CAD: 03brlcad * r42106 10/brlcad/trunk/src/conv/ (3 files in 2 dirs): |
| 00:59.53 | CIA-43 | BRL-CAD: behold the annoyance of -pedantic. believe it or not, even for ISO C++, |
| 00:59.54 | CIA-43 | BRL-CAD: variable sized arrays are but a mere gcc extension. if you want variable-sized, |
| 00:59.55 | CIA-43 | BRL-CAD: you either have to new/delete/malloc/free (bleh) or leverage std::vector |
| 00:59.56 | CIA-43 | BRL-CAD: templatization. the latter is fortunately a trivial declaration tweak and the |
| 00:59.57 | CIA-43 | BRL-CAD: rest should behave accordingly. |
| 00:59.59 | CIA-43 | BRL-CAD: 03starseeker * r42108 10/brlcad/branches/cmake/TODO.cmake: Sigh. Update the TODO list for CMake with more known items. |
| 01:02.34 | *** join/#brlcad crazy_imp (~mj@a89-182-195-57.net-htp.de) | |
| 01:06.40 | CIA-43 | BRL-CAD: 03starseeker * r42109 10/brlcad/branches/cmake/CMakeLists.txt: Want to set to /usr/brlcad as default instead of the CMake default. |
| 01:12.02 | CIA-43 | BRL-CAD: 03brlcad * r42110 10/brlcad/trunk/src/conv/iges/ (g-iges.c main.c usage.c): more string literals that are way too long. convert to usage() functions. |
| 01:12.26 | CIA-43 | BRL-CAD: 03brlcad * r42111 10/brlcad/trunk/src/conv/comgeom/tools.c: curious one, untested. |
| 01:13.56 | CIA-43 | BRL-CAD: 03brlcad * r42112 10/brlcad/trunk/src/vdeck/vdeck.c: more size_t |
| 01:19.21 | CIA-43 | BRL-CAD: 03brlcad * r42113 10/brlcad/trunk/NEWS: john added support for comments as the first line(s) of a .asc file. previously was using title/units as keywords to recognize a .asc file, now it's the first non-comment line that has to have a title/units command. |
| 01:27.21 | CIA-43 | BRL-CAD: 03starseeker * r42114 10/brlcad/branches/cmake/ (45 files in 45 dirs): Generalize the INSTALL_* variables - gives a parent project the chance to do its own setting, in principle, without having to use the BRLCAD specific variables. Not sure how useful it really is, but why not. |
| 01:28.16 | CIA-43 | BRL-CAD: 03starseeker * r42115 10/brlcad/branches/cmake/TODO.cmake: reminder - need to rework SCL CMake logic |
| 01:30.28 | CIA-43 | BRL-CAD: 03starseeker * r42116 10/brlcad/branches/cmake/src/libbu/CMakeLists.txt: Add timer.c to libbu CMakeLists.txt file |
| 01:45.26 | CIA-43 | BRL-CAD: 03starseeker * r42117 10/brlcad/branches/cmake/src/tclscripts/ (hv3/pkgIndex.tcl hv3/tclIndex swidgets/scripts/tclIndex): Add back in some of the tclIndex and pkgIndex files - this should all go when CMake becomes mainline, but for now put them back to avoid difference with trunk. |
| 01:46.27 | CIA-43 | BRL-CAD: 03brlcad * r42118 10/brlcad/trunk/ (6 files in 5 dirs): |
| 01:46.28 | CIA-43 | BRL-CAD: differentiate BU_PTBL_LEN() from BU_PTBL_END() such that the prior is not a |
| 01:46.28 | CIA-43 | BRL-CAD: valid lvalue. this makes it more appropriate for loop testing and can be an |
| 01:46.29 | CIA-43 | BRL-CAD: unsigned type instead of the potentially signed off_t offset type of the |
| 01:46.29 | CIA-43 | BRL-CAD: underlying struct member. update references accordingly. |
| 01:48.48 | CIA-43 | BRL-CAD: 03starseeker * r42119 10/brlcad/branches/cmake/ (37 files in 14 dirs): Update cmake branch to trunk r42052. This is known to be a non-working CMake state, but this sync is being done in stages to avoid some conflicts from the trunk CMakeLists.txt files. |
| 01:55.04 | CIA-43 | BRL-CAD: 03brlcad * r42120 10/brlcad/trunk/src/ (5 files in 4 dirs): more quieting of the compilation |
| 01:58.46 | CIA-43 | BRL-CAD: 03starseeker * r42121 10/brlcad/branches/cmake/ (197 files in 44 dirs): Update cmake branch to trunk r42119 |
| 02:33.16 | CIA-43 | BRL-CAD: 03starseeker * r42122 10/brlcad/branches/cmake/ (7 files in 6 dirs): Update cmake branch to trunk r42120 |
| 02:58.06 | CIA-43 | BRL-CAD: 03starseeker * r42123 10/brlcad/branches/cmake/ (CMakeLists.txt misc/CMake/BRLCAD_Util.cmake): Tweaks - seems to produces better results for make package. |
| 03:12.51 | CIA-43 | BRL-CAD: 03starseeker * r42124 10/brlcad/branches/cmake/misc/CMake/FindTCL.cmake: Add the deprecated Tcl/Tk vars into the handle-standard-args logic - they apparently aren't visible to other projects otherwise. |
| 03:27.43 | CIA-43 | BRL-CAD: 03starseeker * r42125 10/brlcad/branches/cmake/TODO.cmake: Tweaks to settings seem to have CPack behaving better - need to study results a lot more, but at least the bin directory has more than a couple dozen binaries now. |
| 03:29.07 | CIA-43 | BRL-CAD: 03brlcad * r42126 10/brlcad/trunk/src/conv/dxf/dxf-g.c: init vars before testing |
| 03:33.24 | CIA-43 | BRL-CAD: 03brlcad * r42127 10/brlcad/trunk/src/conv/g-acad.c: |
| 03:33.25 | CIA-43 | BRL-CAD: massive restructure in order to fix a bug assuming that variables are accessible |
| 03:33.25 | CIA-43 | BRL-CAD: after a longjmp. quite a chore to quell the warning due to a bug in pre 4.3 |
| 03:33.26 | CIA-43 | BRL-CAD: gcc, but seems to be possible to quiet the warning if we start a new frame (and |
| 03:33.26 | CIA-43 | BRL-CAD: don't access the var/arg after that call). some testing, seems to work. |
| 03:33.55 | CIA-43 | BRL-CAD: 03brlcad * r42128 10/brlcad/trunk/src/conv/Makefile.am: not quite intentional to enable strict in here just yet. -pedantic is a bitch. |
| 04:17.10 | CIA-43 | BRL-CAD: 03brlcad * r42129 10/brlcad/trunk/src/librt/primitives/bot/bot.c: off-by-one copypaste typo. 2, not k. |
| 04:32.48 | CIA-43 | BRL-CAD: 03starseeker * r42130 10/brlcad/branches/cmake/CMakeLists.txt: Tweak generator list, vars for CPack |
| 04:33.57 | CIA-43 | BRL-CAD: 03starseeker * r42131 10/brlcad/branches/cmake/src/other/ (tcl/doc/install_man.cmake.in tk/doc/install_man.cmake.in): the Tcl/Tk man page script was installing to CMAKE_INSTALL_DIR even during make package - make it respect DESTDIR if it's set. |
| 04:56.43 | starseeker | hmm... that won't do either, CPack ignores those man pages as not being from a target |
| 05:03.26 | *** join/#brlcad Stattrav (~Stattrav@122.172.16.143) | |
| 05:03.32 | *** join/#brlcad Stattrav (~Stattrav@unaffiliated/stattrav) | |
| 05:29.12 | CIA-43 | BRL-CAD: 03starseeker * r42132 10/brlcad/branches/cmake/src/other/ (4 files in 2 dirs): Make another stab at the Tcl/Tk man pages, this time doing the generation up front and using FILE(GLOB to grab the results and put them into install targets. |
| 05:35.57 | CIA-43 | BRL-CAD: 03starseeker * r42133 10/brlcad/branches/cmake/src/other/ (tcl/doc/CMakeLists.txt tk/doc/CMakeLists.txt): |
| 05:35.58 | CIA-43 | BRL-CAD: Only do the generation once. If we really want to do this right we need some |
| 05:35.59 | CIA-43 | BRL-CAD: kind of custom commands, targets, output files to signify completed commands, |
| 05:35.59 | CIA-43 | BRL-CAD: etc - not worth it, and this avoids repeated processing we don't need. |
| 05:50.14 | CIA-43 | BRL-CAD: 03starseeker * r42134 10/brlcad/branches/cmake/src/other/tcl/CMakeLists.txt: Tcl link workaround isn't working for make package - need to re-examine. |
| 05:54.07 | CIA-43 | BRL-CAD: 03starseeker * r42135 10/brlcad/branches/cmake/src/other/tcl/doc/CMakeLists.txt: Arrrrgh - make package is still not grabbing these files. May have to go all out with custom commands and targets. |
| 07:26.22 | *** join/#brlcad AlecTaylor (~Tauk@unaffiliated/alectaylor) | |
| 09:54.50 | *** join/#brlcad mafm_ (~mafm@166.Red-83-45-72.dynamicIP.rima-tde.net) | |
| 11:38.09 | DaveLo | Mernin! Anyone at work yet? |
| 13:36.40 | CIA-43 | BRL-CAD: 03d_rossberg * r42136 10/brlcad/trunk/misc/win32-msvc/Dll/BrlcadCore.def: added bu_vls_trunc2() for asc2g conversion program |
| 13:40.40 | starseeker | hmm... KDE on Windows |
| 13:41.06 | starseeker | is intrigued, but hesitates to mess with his Windows partition too much because it would be hard to fix |
| 13:46.02 | starseeker | huh - kinda looks like bzflag with some enhancements: http://www.ubuntugamer.com/2011/01/zero-ballistics-a-rather-addictive-native-3d-tank-shooter/ |
| 13:51.15 | *** join/#brlcad mafm (~mafm@166.Red-83-45-72.dynamicIP.rima-tde.net) | |
| 13:52.13 | CIA-43 | BRL-CAD: 03starseeker * r42137 10/brlcad/branches/cmake/src/other/ (tcl/doc/CMakeLists.txt tk/doc/CMakeLists.txt): Ah, was shooting myself in the foot. We only GENERATE the man pages once, but we add them to the INSTALL list every time cmake is run. |
| 14:04.21 | *** join/#brlcad cjdevlin (~devlin@d118-75-252-178.try.wideopenwest.com) | |
| 14:15.09 | *** join/#brlcad d_rossberg (~rossberg@BZ.BZFLAG.BZ) | |
| 14:36.21 | CIA-43 | BRL-CAD: 03starseeker * r42138 10/brlcad/branches/cmake/CMakeLists.txt: |
| 14:36.22 | CIA-43 | BRL-CAD: Small change, but important for CPack packages - don't use absolute dirs as |
| 14:36.22 | CIA-43 | BRL-CAD: targets for install, which allows CPack to put things in the package without |
| 14:36.23 | CIA-43 | BRL-CAD: hard-coding the prefix (in essence building a subdirectory structure inside the |
| 14:36.23 | CIA-43 | BRL-CAD: archive |
| 14:55.59 | CIA-43 | BRL-CAD: 03starseeker * r42139 10/brlcad/branches/cmake/src/ (7 files in 7 dirs): |
| 14:56.00 | CIA-43 | BRL-CAD: Strip out more absolute paths for installs. The src/other/step directory isn't |
| 14:56.00 | CIA-43 | BRL-CAD: with the program yet, but that needs major cleanup anyway so leave it for now. |
| 14:56.01 | CIA-43 | BRL-CAD: Probably shouldn't be hardcoding bin and lib in as DESTINATIONS, may want to do |
| 14:56.01 | CIA-43 | BRL-CAD: a sweep for that too. |
| 15:02.21 | brlcad | starseeker: have you looked at tom's crash? have a question on tedit for you |
| 15:03.17 | brlcad | what's the intent around line 986 where it calls: editor = bu_which(EMACS_EDITOR); if (!strcmp(editor, bu_which(EMACS_EDITOR)) ... |
| 15:03.54 | d_rossberg | i could reproduce the crash, now i'm looking for a work-around |
| 15:04.15 | brlcad | d_rossberg: I have a simplistic fix for the crash that at least avoid strcmp() |
| 15:05.53 | CIA-43 | BRL-CAD: 03brlcad * r42140 10/brlcad/trunk/src/mged/tedit.c: |
| 15:05.54 | CIA-43 | BRL-CAD: strcmp() doesn't like NULL and bu_which() may produce NULL, so don't feed the |
| 15:05.54 | CIA-43 | BRL-CAD: output of the latter directly into the prior. should at least avoid the |
| 15:05.55 | CIA-43 | BRL-CAD: specific crash reported by tom browder to the devel mailing list, albeit |
| 15:05.55 | CIA-43 | BRL-CAD: probably not the full fix needed. |
| 15:09.27 | starseeker | brlcad: ah, thanks - I neglected to check if strcmp would handle null. Looks like it's platform dependent, and therefore not to be relied on |
| 15:11.03 | brlcad | there's not many stdc calls that can be trusted to accept null, even ones that are documented to accept null |
| 15:11.17 | d_rossberg | on Windows strcmp(0) will crash |
| 15:13.02 | CIA-43 | BRL-CAD: 03starseeker * r42141 10/brlcad/branches/cmake/src/archer/CMakeLists.txt: Archer's CMakeLists.txt file needs a lot of love to make it work in the build dir - these are just the first tweaks, lots more will be needed. |
| 15:13.02 | starseeker | sigh |
| 15:13.25 | starseeker | launching an editor from C code really seems to be a pain |
| 15:13.32 | brlcad | I believe posix strcmp() is undefined if passed a null anyways, so crashing is acceptable behavior |
| 15:13.45 | d_rossberg | whot do you think about a bu_strcmp()? |
| 15:14.27 | brlcad | you mean adding one? |
| 15:14.45 | d_rossberg | yes |
| 15:15.16 | starseeker | d_rossberg: (bty - more to be done, but hopefully CMAKE_INSTALL_PREFIX will behave more normally now and you won't need BRLCAD_PREFIX anymore." |
| 15:15.41 | starseeker | leftover garbage from early in the learning process |
| 15:15.44 | d_rossberg | strcmp() is really annoying |
| 15:17.49 | starseeker | gets ready to head in... |
| 15:17.56 | CIA-43 | BRL-CAD: 03starseeker * r42142 10/brlcad/branches/cmake/TODO.cmake: Add note about archer to TODO.cmake |
| 15:18.08 | d_rossberg | starseeker: i've still the problem of huge rebuilds ... |
| 15:21.25 | brlcad | d_rossberg: bu_strcmp() and/or bu_strlcmp() would be good additions to make given there is already bu_str(lcat|lcpy|dup) |
| 15:21.44 | brlcad | starseeker: any insight on that question? |
| 15:22.15 | d_rossberg | brlcad: i'm already working on it |
| 15:22.27 | starseeker | brlcad: the strcmp or the rebuilds? |
| 15:22.35 | brlcad | what's the intent around line 986 where it calls: editor = bu_which(EMACS_EDITOR); if (!strcmp(editor, bu_which(EMACS_EDITOR)) ... |
| 15:22.46 | brlcad | now line 904 or something |
| 15:23.18 | brlcad | it sets editor to the path to emacs, then checks if editor is .. the path to emacs |
| 15:23.33 | starseeker | looks... |
| 15:24.16 | brlcad | I added the "(editor &&" part just to check the result from bu_which .. wasn't there before |
| 15:24.24 | brlcad | *just* added it in the last commit |
| 15:24.33 | brlcad | d_rossberg: okay, cool |
| 15:26.52 | starseeker | I believe what was going on there was a check (once in classic mode) to see if any of the known editor configurations that would work in classic mode were viable |
| 15:27.30 | brlcad | I think I see that, but those two specific lines don't make sense to me |
| 15:28.22 | starseeker | um. |
| 15:28.28 | starseeker | yeah, not sure what was going on there |
| 15:28.40 | starseeker | that strcmp does look kinda pointless |
| 15:29.16 | starseeker | I think you have the right answer |
| 15:29.20 | brlcad | well what I'm reading is that IF you have emacs, then all of the other editor tests will fail |
| 15:29.26 | brlcad | which maybe was the intent |
| 15:29.31 | d_rossberg | brlcad: there is no strlcmp |
| 15:29.41 | brlcad | d_rossberg: I meant to write one :) |
| 15:30.13 | starseeker | brlcad: probably it was something like that |
| 15:30.21 | brlcad | with similar intent of the strlcpy functions, strlcmp checks for null, maybe has a length specifier |
| 15:30.33 | d_rossberg | there are strl*() functions for writing on buffers only |
| 15:31.47 | brlcad | okay |
| 15:32.19 | brlcad | fair enough, it was just a thought for consistency with our own bu functions, but matching the replacement is probably more important |
| 15:32.44 | ``Erik | now that's a lot of breakage |
| 15:33.13 | starseeker | brlcad: I think I may have some rather convoluted logic there - I'm not sure what the whole idea was with the count thing |
| 15:33.21 | brlcad | I could see bu also providing a BU_STREQ() macro so that all of the equality testing could be simplified consistent |
| 15:34.12 | brlcad | starseeker: yeah, that is pretty funky ... search for everything, then if you found anything, search for something |
| 15:34.46 | starseeker | oh, wait |
| 15:35.11 | starseeker | I think I might have been looking to see if a preset editor from above was a potential bad case for classic mode |
| 15:35.43 | starseeker | looks at the previous revision of the code |
| 15:36.05 | brlcad | hm, maybe if(BU_EQUALS(str1, str2)) or if (BU_EQUAL_STR(str1, str2)) |
| 15:36.55 | starseeker | yeah, that was it |
| 15:36.57 | ``Erik | using vls's? |
| 15:37.07 | brlcad | there are approximately 1418 calls to strcmp() in the code |
| 15:38.12 | starseeker | brlcad: the idea was to check the editor variable to spot cases which might be a problem, and if it WAS a problem case then force-feed it something known to be safe |
| 15:38.58 | starseeker | so it blew up due to the bu_which finding nothing to compare editor TO |
| 15:39.17 | brlcad | right |
| 15:39.36 | brlcad | so then my fix should, in theory, work and fall through correctly |
| 15:39.38 | starseeker | so your fix is actually correct |
| 15:39.41 | starseeker | yes :-) |
| 15:39.59 | starseeker | shudders in memory - that was a long night sorting all that logic out |
| 15:41.08 | brlcad | those first two lines inside the if (count > 0) block still don't make sense though |
| 15:41.39 | brlcad | because editor = bu_which(EMACS_EDITOR) will have unlikely changed between the first and second lines.. :) |
| 15:41.47 | starseeker | right :-) |
| 15:42.20 | starseeker | I think that was brain-overload coding - it functioned, so go with it |
| 15:42.28 | starseeker | removes the stray strcmp |
| 15:42.49 | brlcad | did you actually see "editor[0] == '\0'" ? |
| 15:43.02 | starseeker | uh - don't recall |
| 15:43.04 | brlcad | bu_which() certainly shouldn't be returning that -- could even test for that |
| 15:43.11 | brlcad | (in bu_which()) |
| 15:43.25 | starseeker | probably not - I think it was more stupidity on my part |
| 15:46.16 | CIA-43 | BRL-CAD: 03brlcad * r42143 10/brlcad/trunk/src/libbu/ (whereis.c which.c): add code to make sure, even though it's a very unlikely event, that we never return an empty string. |
| 15:46.18 | CIA-43 | BRL-CAD: 03starseeker * r42144 10/brlcad/trunk/src/mged/tedit.c: |
| 15:46.18 | CIA-43 | BRL-CAD: We can trust bu_wish not to return \0, so we don't need that check there. May |
| 15:46.24 | CIA-43 | BRL-CAD: not need it anywhere, but we are getting returns from other functions early on |
| 15:46.24 | CIA-43 | BRL-CAD: so would need more careful checking. Also remove the useless strcmp for editor |
| 15:46.24 | CIA-43 | BRL-CAD: just after setting it to emacs. |
| 15:47.05 | starseeker | brlcad: uh, it's close to a certainty that my brain said "check for an empty char array here" and did something colossally stupid - I doubt bu_which is at fault |
| 15:48.10 | brlcad | starseeker: it's still a reasonable "guarantee" that bu_which() should be able to say, NULL if not found, non-NULL non-empty if found |
| 15:48.17 | starseeker | has been doing too much build logic... s/bu_wish/bu_which |
| 15:48.35 | starseeker | cool |
| 15:52.34 | starseeker | does head in this time... |
| 15:52.49 | d_rossberg | i like the NULL as a return value because i haven't to provide memory for it |
| 15:57.04 | CIA-43 | BRL-CAD: 03brlcad * r42145 10/brlcad/trunk/src/conv/iges/iges.c: make sure initd before use, compiler doesn't see the bu_bomb(). |
| 15:58.29 | CIA-43 | BRL-CAD: 03brlcad * r42146 10/brlcad/trunk/src/conv/ (euclid/g-euclid.c iges/g-iges.c): restructure exception handling into try/catch form |
| 16:07.55 | CIA-43 | BRL-CAD: 03d_rossberg * r42147 10/brlcad/trunk/ (include/bu.h src/libbu/str.c): |
| 16:07.56 | CIA-43 | BRL-CAD: introduced bu_strcmp() with a more graceful string comparison as strcmp() does |
| 16:07.56 | CIA-43 | BRL-CAD: "" and NULL are considered as equal |
| 16:08.08 | d_rossberg | i've always wanted to do that |
| 16:25.11 | CIA-43 | BRL-CAD: 03d_rossberg * r42148 10/brlcad/trunk/include/bu.h: |
| 16:25.12 | CIA-43 | BRL-CAD: the BU_STR_EMPTY() macro tests a string for emptiness ("" or NULL) |
| 16:25.13 | CIA-43 | BRL-CAD: its result is either true or false |
| 16:43.19 | CIA-43 | BRL-CAD: 03brlcad * r42149 10/brlcad/trunk/src/conv/iges/ (91 files): cleanup of the old and new iges codes. ws, consistency, de-k&r, indent, authorship, etc. |
| 16:47.55 | brlcad | starseeker: I take it back, the old iges code is so much more extensive than the new that it'd be just ridiculous to dump it for the new one |
| 16:48.25 | brlcad | the new one does have some nice stucture to it, but it's so devoid of implementation that it seems better just because there is so little complexity (or value) involved |
| 16:56.52 | CIA-43 | BRL-CAD: 03brlcad * r42150 10/brlcad/trunk/include/bu.h: (log message trimmed) |
| 16:56.53 | CIA-43 | BRL-CAD: follow suit and provide another BU_STR_EQUAL() macro for comparing two strings |
| 16:56.53 | CIA-43 | BRL-CAD: for equality. this is similar to the STREQ recommendation by the ibm |
| 16:56.54 | CIA-43 | BRL-CAD: developerworks best practices article |
| 16:56.54 | CIA-43 | BRL-CAD: (http://www.ibm.com/developerworks/aix/library/au-hook_duttaC.html) so that |
| 16:56.55 | CIA-43 | BRL-CAD: developers can use equality as a truthfulness return value consistently. there |
| 16:57.09 | CIA-43 | BRL-CAD: are approximately 1400 present uses of strcmp() in BRL-CAD that can use this |
| 17:32.49 | starseeker | brlcad: so... gradually refactor the old code to use better structure and convert to the new nurbs? |
| 17:33.10 | starseeker | or I suppose do so as part of merging it into libgcv |
| 17:40.22 | CIA-43 | BRL-CAD: 03starseeker * r42151 10/brlcad/branches/cmake/ (102 files in 9 dirs): Update cmake branch to trunk r41250 |
| 17:45.29 | brlcad | starseeker: there is always possibility for better structure, particularly the bigger, older, and more complex code becomes ... so that's a given |
| 17:46.01 | brlcad | but yeah, merge in capability for new nurbs would probably be better -- it'll just be more costly up front to learn the existing code |
| 17:48.03 | brlcad | unrelated topic, the compiler is reporting that it cannot inline the opennurbs_ext BBNode bounding box routine (GetBBox()), so there is possibly some significant performance to be gained by fixing that |
| 18:06.55 | CIA-43 | BRL-CAD: 03starseeker * r42152 10/brlcad/branches/cmake/src/conv/iges/revolve.c: PI is coming up as undefined on OSX - use M_PI |
| 18:10.58 | CIA-43 | BRL-CAD: 03brlcad * r42153 10/brlcad/trunk/src/librt/opennurbs_ext.h: force inlining on the bounding box routines since gcc (4.0.1) complains that it cannot without increasing an inline limit. remove inline from destructors (gcc is similiarly complaining that it cannot). |
| 18:11.50 | CIA-43 | BRL-CAD: 03brlcad * r42154 10/brlcad/trunk/src/librt/ (bool.c comb/comb.c): sign matching |
| 18:32.42 | CIA-43 | BRL-CAD: 03erikgreenwald * r42155 10/brlcad/trunk/src/libbn/ (bntester.c poly.c tabdata.c): fix various warnings |
| 18:35.08 | CIA-43 | BRL-CAD: 03starseeker * r42156 10/brlcad/branches/cmake/src/other/tcl/ (4 files in 4 dirs): Start the process of taking things out of -D defines and putting them in config headers. |
| 18:48.27 | CIA-43 | BRL-CAD: 03starseeker * r42157 10/brlcad/branches/cmake/src/other/ (5 files in 5 dirs): |
| 18:48.27 | CIA-43 | BRL-CAD: more -D elimination - a lot of these defines probably aren't even needed in |
| 18:48.28 | CIA-43 | BRL-CAD: CMake builds, as they don't seem to be used by the C code and CMake is handling |
| 18:48.29 | CIA-43 | BRL-CAD: the generation of whatever config files are needed itself... for the ones that |
| 18:48.33 | CIA-43 | BRL-CAD: are, the only code change is to include the generated header. Definitely more |
| 18:48.33 | CIA-43 | BRL-CAD: cleanup to do on these to reduce the build logic to the functioning minimum. |
| 18:53.19 | CIA-43 | BRL-CAD: 03starseeker * r42158 10/brlcad/branches/cmake/CMakeLists.txt: Let's try a space on Windows again and see if those defines fixed it... |
| 18:59.41 | CIA-43 | BRL-CAD: 03brlcad * r42159 10/brlcad/trunk/src/librt/primitives/nmg/nmg.c: size_t |
| 19:00.09 | CIA-43 | BRL-CAD: 03brlcad * r42160 10/brlcad/trunk/src/librt/primitives/brep/brep.cpp: VMOVE before we print the value. |
| 19:01.24 | CIA-43 | BRL-CAD: 03brlcad * r42161 10/brlcad/trunk/src/librt/primitives/nmg/nmg_brep.cpp: init max_pt too, quell warning |
| 19:06.31 | CIA-43 | BRL-CAD: 03brlcad * r42162 10/brlcad/trunk/ (8 files in 2 dirs): make a slew of other object count struct data members be size_t instead of long and int so that can be properly unsigned and higher bound. match signedness in librt accordingly. |
| 19:14.49 | CIA-43 | BRL-CAD: 03brlcad * r42163 10/brlcad/trunk/src/ (9 files in 2 dirs): another BU_PTBL_LEN caller, size_t it up. |
| 19:17.10 | CIA-43 | BRL-CAD: 03brlcad * r42164 10/brlcad/trunk/include/bu.h: just call bu_strcmpm() directly instead of macro. didn't get the preproc syntax right anyways. |
| 19:26.35 | brlcad | is excited that we're SO CLOSE to a strict clean build! |
| 19:29.16 | brlcad | improved security, maintainability, conformance/verification/validation, .. yum |
| 19:34.56 | brlcad | envisions a day where the source code is completely 100% lintian free with best practices enforced across the entire 1M+ body of code |
| 19:37.36 | CIA-43 | BRL-CAD: 03brlcad * r42165 10/brlcad/trunk/src/conv/iges/revolve.c: didn't starseeker already fix this? M_PI is the new coke. |
| 19:39.50 | CIA-43 | BRL-CAD: 03brlcad * r42166 10/brlcad/trunk/src/conv/iges/iges.c: multicharacter string constants are not valid to cpp, so move the literal to a define in order to catch future auto-expansions |
| 20:23.25 | *** join/#brlcad ibot (~ibot@198.60.114.207) | |
| 20:23.25 | *** topic/#brlcad is BRL-CAD Open Source Solid Modeling || http://brlcad.org || http://sf.net/projects/brlcad || #brlcad logs: http://ibot.rikers.org/%23brlcad/ || BRL-CAD release 7.18.0 is posted (20101209) || Happy Open Source Anniversary 2010-12-21 !!! Six years... | |
| 20:28.16 | CIA-43 | BRL-CAD: 03brlcad * r42170 10/brlcad/trunk/src/fb/pp-fb.c: init to zero pixels |
| 20:28.17 | CIA-43 | BRL-CAD: 03starseeker * r42169 10/brlcad/branches/cmake/CMakeLists.txt: Interesting - the -c option causes the time delta compile to fail. |
| 20:30.20 | CIA-43 | BRL-CAD: 03brlcad * r42173 10/brlcad/trunk/src/util/ (pc_test.c pixborder.c pixcount.c pixdsplit.c): more unset before use insanity. |
| 20:30.49 | CIA-43 | BRL-CAD: 03brlcad * r42174 10/brlcad/trunk/bench/pixcmp.c: test of conversion to BU_STR_EQUAL() instead of directly calling strcmp(). |
| 20:31.23 | DaveLo | getting an error in arbn.c: |
| 20:31.37 | DaveLo | primitives/arbn/arbn.c: In function ?rt_arbn_describe?: |
| 20:31.37 | DaveLo | primitives/arbn/arbn.c:1026: error: format ?%lu? expects type ?long unsigned int?, but argument 3 has type ?size_t? |
| 20:31.40 | DaveLo | primitives/arbn/arbn.c:1037: error: format ?%lu? expects type ?long unsigned int?, but argument 3 has type ?size_t? |
| 20:32.32 | ``Erik | lots of those |
| 20:32.33 | ``Erik | lots and lot |
| 20:32.34 | ``Erik | s |
| 20:33.48 | ``Erik | thinks he's about to get very involved with GS O>o |
| 20:33.54 | CIA-43 | BRL-CAD: 03erikgreenwald * r42175 10/brlcad/trunk/src/adrt/ (27 files in 3 dirs): warning quellage. formatting fixes. |
| 20:34.19 | DaveLo | what makes you say that? |
| 20:35.07 | *** join/#brlcad ibot (~ibot@rikers.org) | |
| 20:35.07 | *** topic/#brlcad is BRL-CAD Open Source Solid Modeling || http://brlcad.org || http://sf.net/projects/brlcad || #brlcad logs: http://ibot.rikers.org/%23brlcad/ || BRL-CAD release 7.18.0 is posted (20101209) || Happy Open Source Anniversary 2010-12-21 !!! Six years... | |
| 20:35.10 | ``Erik | (k keeps building after failure) |
| 20:35.23 | ``Erik | or didja mean GS? |
| 20:35.58 | DaveLo | huh? |
| 20:36.21 | starseeker | what made him say what? |
| 20:36.33 | DaveLo | whos a what? |
| 20:36.40 | ``Erik | forshizzle? |
| 20:36.48 | DaveLo | forrizzle! |
| 20:40.58 | DaveLo | FYI: I was updating the repos on my Ubuntutututu box and ran into some size_t compile errors in brlcad |
| 20:41.27 | DaveLo | figured I'd mention it since brlcad is in the mist of some massive size_t work |
| 21:13.20 | CIA-43 | BRL-CAD: 03starseeker * r42176 10/brlcad/branches/cmake/src/tclscripts/CMakeLists.txt: Remove old files before making new ones... |
| 21:37.00 | CIA-43 | BRL-CAD: 03starseeker * r42177 10/brlcad/branches/cmake/src/other/ (6 files in 6 dirs): Remove all the -c flags from the Tcl/Tk builds |
| 21:40.31 | *** join/#brlcad Zaebos (~irc@pd95b7f5e.dip0.t-ipconnect.de) | |
| 21:56.14 | brlcad | DaveLo: I've got a clean build on mac and linux, but size_t is going to be pretty sensitive to platform differences so some failures undoubtedly need weeding out |
| 21:56.29 | brlcad | should be trivial fixes, at least |
| 21:57.00 | brlcad | DaveLo: and --enable-warnings is now the default (as of yesterday) |
| 22:08.21 | CIA-43 | BRL-CAD: 03starseeker * r42178 10/brlcad/branches/cmake/CMakeLists.txt: For whatever reason, CMAKE_C_FLAGS is upsetting VC++ 2010 - comment it out |
| 22:15.44 | CIA-43 | BRL-CAD: 03erikgreenwald * r42179 10/brlcad/trunk/src/libfb/if_ogl.c: signed/unsigned comparison fixes |
| 22:41.39 | ``Erik | brlcad: if'n ya get bored, http://brlcad.org/~erik/r42179/ :D *heads home* |
| 22:49.49 | brlcad | heh |
| 22:49.52 | CIA-43 | BRL-CAD: 03brlcad * r42180 10/brlcad/trunk/src/libbu/CMakeLists.txt: sync timetester to cmake build for distcheck |
| 22:50.41 | starseeker | growls... CMake build doesn't run as of 42152 |
| 22:50.55 | starseeker | seems to run at 42150, confirming that |
| 22:51.15 | starseeker | what could possibly have changed... |
| 22:52.47 | brlcad | i've done so much fast typing for over a week now that my rsi is starting to act up |
| 22:53.08 | starseeker | winces - that's not good |
| 22:54.40 | brlcad | not too bad just yet, just having to take more breaks |
| 22:55.04 | brlcad | editing thousands of files will do that |
| 22:56.01 | brlcad | starseeker: i'd be cautious on any failure -- those size_t changes could easily have (and undoubtedly have in some places) caused a bug to get injected somewhere/anywhere |
| 22:56.19 | brlcad | along with the changes before I got on the size_t rampage even |
| 22:56.49 | starseeker | nods - I'm trying to isolate when it happend |
| 22:57.23 | starseeker | will cheerfully revert the source of cmake branch to a known good state and work on that, if he can find such a state |
| 23:02.53 | starseeker | OK, NOT working at 42150 |
| 23:09.52 | starseeker | 42139 working |
| 23:09.58 | starseeker | did that archer change mess things up somehow??? |
| 23:25.10 | CIA-43 | BRL-CAD: 03starseeker * r42181 10/brlcad/branches/cmake/src/archer/CMakeLists.txt: Hmm - this makes mged REALLY unhappy, so don't do it... |
| 23:26.46 | CIA-43 | BRL-CAD: 03brlcad * r42182 10/brlcad/trunk/src/mged/hideline.c: protect from peculiar setjmp() use in here by making the variables set before and accessed afterwards as static. candidate for removal. |
| 23:29.17 | CIA-43 | BRL-CAD: 03brlcad * r42183 10/brlcad/trunk/src/mged/dodraw.c: |
| 23:29.17 | CIA-43 | BRL-CAD: one of the more complicated ways to handle BU_SETJUMPing so that variables set |
| 23:29.18 | CIA-43 | BRL-CAD: before the jump and accessed afterwards do not have their values clobbered when |
| 23:29.18 | CIA-43 | BRL-CAD: a jump occurs. solution is to pull just exactly the try/catch code out into |
| 23:29.19 | CIA-43 | BRL-CAD: their own functions so there are no variables in that frame that might be |
| 23:29.19 | CIA-43 | BRL-CAD: clobbered in the first place. this is done twice here. |
| 23:29.52 | CIA-43 | BRL-CAD: 03brlcad * r42184 10/brlcad/trunk/src/mged/cmd.c: size_t upconvert argc |
| 23:30.17 | starseeker | O.o - that's some scary sounding logic - why are we need jumps like that, speed? |
| 23:30.51 | starseeker | breaths a sigh of relief - MGED runs again |
| 23:39.13 | CIA-43 | BRL-CAD: 03brlcad * r42185 10/brlcad/trunk/src/mged/rtif.c: wrong comment to the wrong file. protect from peculiar setjmp() use in here by making the variables set before and accessed afterwards as static. candidate for removal. |
| 23:39.56 | CIA-43 | BRL-CAD: 03brlcad * r42186 10/brlcad/trunk/src/mged/mged.c: make sure we have an out if we don't have an out. |
| 23:40.51 | CIA-43 | BRL-CAD: 03brlcad * r42187 10/brlcad/trunk/src/mged/ (cad_parea.c chgview.c clone.c): init vars before they're used, especially when they're only set within conditionals. |
| 23:42.19 | starseeker | heh 42186 sounds like Yogi Berra |
| 23:46.32 | CIA-43 | BRL-CAD: 03brlcad * r42188 10/brlcad/trunk/src/mged/cad_boundp.c: one more needing to be initialized |
| 23:47.46 | brlcad | starseeker: jumps are the old school way to perform exception handling |
| 23:48.11 | brlcad | c++'s exception handling was originally implemented using setjmp/longjmp and macros |
| 23:48.37 | brlcad | you just have to know what you're doing and what happens when a jump occurs |
| 23:49.02 | brlcad | some of our code was making assumptions which work in practice, but aren't guaranteed |
| 23:49.20 | starseeker | ah |
| 23:49.50 | brlcad | basically, IF a jump happens, variables are reset back to their state when the jumppoint was *set* .. so if you modify them after the set and then jump, their values are clobbered |
| 23:50.17 | brlcad | most of the time, that's perfectly fine |
| 23:51.03 | brlcad | and whether it's fine or not, doing the "wrap the try/catch" in a function trick makes the problem pretty much moot because there are no longer any stack variables getting clobbered |
| 23:51.52 | starseeker | nods |
| 23:53.17 | brlcad | I want to rewrite our macros so we're not setting and unsetting jumps, instead providing BU_TRY/BU_CATCH macros |
| 23:53.39 | starseeker | how many places in the code will that touch? |
| 23:56.28 | starseeker | worries about brlcad's wrists |
| 23:56.47 | brlcad | I don't remember how many places, not too many |
| 23:56.54 | starseeker | cool |
| 23:56.58 | brlcad | unfortunately not really scriptable though :) |
| 23:57.03 | starseeker | heh |
| 23:57.13 | brlcad | just scan on BU_SETJUMP though and you'll find them all |
| 23:58.32 | starseeker | popular in the convertors |
| 23:58.52 | brlcad | yep, nmg stuff throws exceptions as part of normal business |
| 23:59.43 | brlcad | BU_SETJUMP is the way to catch a bu_bomb() so it doesn't actually exit |