Activity for Cliff Yapp

  • Cliff Yapp Cliff Yapp committed [r70923]

    Make a stab at detecting when we edited a tag. Initial results look plausible.

  • Cliff Yapp Cliff Yapp committed [r70922]

    Few more setup notes, naming consistency

  • Cliff Yapp Cliff Yapp committed [r70921]

    comment out sleep (using for debugging to make it easier to inspect the .g file state.)

  • Cliff Yapp Cliff Yapp committed [r70920]

    Start stubbing in a test for datums. Demonstrates using the libbu temp file mechanism to get a file and turn it into a .g file.

  • Cliff Yapp Cliff Yapp committed [r70919]

    Be explicit about src/other building using the same 32/64 bit flags as the parent build.

  • Cliff Yapp Cliff Yapp committed [r70916]

    pack results into a struct as well - we've got at least two more types of commit we want to try and characterize - branch creation and edits in tags

  • Cliff Yapp Cliff Yapp committed [r70913]

    Add a configure option to disable GCT building

  • Cliff Yapp Cliff Yapp committed [r70911]

    fix output of help menu

  • Cliff Yapp Cliff Yapp committed [r70909]

    don't quit after a successful debug value setting

  • Cliff Yapp Cliff Yapp committed [r70905]

    plot3-dm uses bu_brlcad_data, so it should bu_setprogname

  • Cliff Yapp Cliff Yapp committed [r70902]

    Forgot bu_setprogname for gcv

  • Cliff Yapp Cliff Yapp committed [r70899]

    CPACK, not CMAKE

  • Cliff Yapp Cliff Yapp committed [r70898]

    Have default NSIS install directory use BRL-CAD, not BRLCAD

  • Cliff Yapp Cliff Yapp committed [r70897]

    fix distcheck

  • Cliff Yapp Cliff Yapp committed [r70896]

    gcv tries to support an Imagemagick '{format}:' prefix syntax for specifying file type, but Windows makes this problematic with its {drive letter}: style of path. Because of this path style being a valid path specifier, we can't support single letter format prefixes without introducing ambiguity. Ironically, the only CAD mime type with a single letter mapping we have defined currently (i.e. the only specification this might actually break) is the .g file. So, as an alternative, we add an a...

  • Cliff Yapp Cliff Yapp committed [r70895]

    reorganize

  • Cliff Yapp Cliff Yapp committed [r70894]

    Work on spotting branch deletes

  • Cliff Yapp Cliff Yapp committed [r70893]

    Attempts to fast-import stream svn commits on top of the cvs->git conversion using svn2git or git-svn-fast-import haven't worked very well so far... probably still possible in theory but not sure how far I'd end up having to dig. For now start doing some experiments on the 'manual replay' scenario to see if that would be simpler to figure out - slow and working is fine for this, as long as it works...

  • Cliff Yapp Cliff Yapp committed [r70892]

    Extract log message for revision (assumes repo clone in addition to dump file, and working svn and xsltproc commands).

  • Cliff Yapp Cliff Yapp committed [r70891]

    Narrow down to files specifically, filter for just the brlcad part of the repository (i.e. don't report paths in rt^3), capture and report the specific paths moved that are reporting as problematic for each commit.

  • Cliff Yapp Cliff Yapp committed [r70890]

    Don't want to skip reporting the last commit, if it also satisfies the criteria.

  • Cliff Yapp Cliff Yapp committed [r70889]

    Thanks to tacit on svn-dev for the hint on what to look for - write a little program to look for SVN commits in a dump file that both moved the file and changed its contents at the same time. Catches a small test case successfully, but needs more testing.

  • Cliff Yapp Cliff Yapp committed [r70888]

    Technically, this may be user visible. Previous NIRT behavior when reading in double floating point numbers used an internal routine that sometimes resulted in an accumulated error (observed in some quick testing to be on the order of 1e-10). While this was insignificant in 'most' use cases, when trying to reproduce an exact ray it could be an issue. Switched to using the standard C library function strtod for string to double conversion, which does a better job.

  • Cliff Yapp Cliff Yapp committed [r70887]

    Add probable account-map id

  • Cliff Yapp Cliff Yapp committed [r70886]

    The INFINITY assignments in shoot.c lines 993-1025 reach the rt_in_rpp calculations, and in cases where these values are multiplied by 0 a multiplication using an actual C99 inf results in a nan that causes all subsequent comparisons to return false. However, when falling back to our C89 compatibility definitions for INFINITY we get typical numerical behavior (0*INFINITY == 0, not nan) which results in different behavior of the comparisons. For specific grazing cases, this results in differ...

  • Cliff Yapp Cliff Yapp committed [r70885]

    It turns out str_dbl was calculating the floating point value by doing an accumulation sum and multiplication, which can result in an observed accumulated numerical delta on the order of 1e-10 between str_dbl and the standard function strtod. The standard function is unsurprisingly correct (comparing it to the input string buffer) so the new behavior actually fixes a bug in nirt. However, using any typical units a delta of 1e-10 is way below BN_TOL_DIST. An argument might be made that the o...

  • Cliff Yapp Cliff Yapp committed [r70884]

    Checkpoint a script working on methods to replay the svn commit history into git - another possible conversion alternative.

  • Cliff Yapp Cliff Yapp committed [r70883]

    Add a convenience tool and some notes about making commit msg md5sum -> revision maps.

  • Cliff Yapp Cliff Yapp committed [r70882]

    I don't think terra.dsp is intended to be treated as a txt file?

  • Cliff Yapp Cliff Yapp committed [r70877]

    Pull in an update from upstream svn2git

  • Cliff Yapp Cliff Yapp committed [r70876]

    ws

  • Cliff Yapp Cliff Yapp committed [r70875]

    Add an ASCII translation and slight update of the spreadsheet Nick and Cliff worked on specifying how search path modifiers should work.

  • Cliff Yapp Cliff Yapp committed [r70871]

    Make a stab at upgrading to 2016 version of pstdint.h header file from http://www.azillionmonkeys.com/qed/pstdint.h, with some of the BRL-CAD changes merged in. Needs testing.

  • Cliff Yapp Cliff Yapp committed [r70870]

    Don't assume ifp calloc always succeeds

  • Cliff Yapp Cliff Yapp committed [r70869]

    Put back the default logic for restricting the configs to Debug and Release, but with an easy way to turn it off at need.

  • Cliff Yapp Cliff Yapp committed [r70867]

    Fix for Ubuntu + gcc7

  • Cliff Yapp Cliff Yapp committed [r70866]

    C++98 doesn't like z - just comment out log message for now

  • Cliff Yapp Cliff Yapp committed [r70865]

    Shouldn't need this line - the output file is explicitly a target output.

  • Cliff Yapp Cliff Yapp committed [r70864]

    Ah, output path wasn't the issue. ninja doesn't do ADDITIONAL_MAKE_CLEAN_FILES, so need DISTCLEAN here.

  • Cliff Yapp Cliff Yapp committed [r70863]

    Put the log in CMAKE_CURRENT_BUILD_DIR_SCRIPT

  • Cliff Yapp Cliff Yapp committed [r70862]

    For whatever reason, Release building mode in MSVC doesn't seem to tolerate bu_malloc/bu_free in tc.c. Trying to use it results in btclsh and asc2g executions hanging indefinitely during the build.

  • Cliff Yapp Cliff Yapp committed [r70861]

    tweaks

  • Cliff Yapp Cliff Yapp committed [r70860]

    Use bio.h for windows.h

  • Cliff Yapp Cliff Yapp committed [r70859]

    Move the clockid_t test in with the other flag setting functions

  • Cliff Yapp Cliff Yapp committed [r70858]

    tcl.cmake needs to include this itself in stand-alone building

  • Cliff Yapp Cliff Yapp committed [r70857]

    CMake build doesn't like this bit...

  • Cliff Yapp Cliff Yapp committed [r70856]

    More tweaking of the standard setting logic

  • Cliff Yapp Cliff Yapp committed [r70855]

    Pull the latest core-8-5-branch updates from Tcl/Tk (via the github mirror). Some patches have been merged post 8.5.19.

  • Cliff Yapp Cliff Yapp committed [r70854]

    Try actually unsetting the standard vars in src/other, rather than setting them to something different.

  • Cliff Yapp Cliff Yapp committed [r70852]

    Look for all the whitespace chars, per isspace - http://www.cplusplus.com/reference/cctype/isspace/

  • Cliff Yapp Cliff Yapp committed [r70850]

    The way MSVC works, Debug and Release aren't actually enough.

  • Cliff Yapp Cliff Yapp committed [r70849]

    length, not start/end positions

  • Cliff Yapp Cliff Yapp committed [r70842]

    build system is not Bourne Shell Script anymore...

  • Cliff Yapp Cliff Yapp committed [r70841]

    Couple stray fixes for MSVC

  • Cliff Yapp Cliff Yapp committed [r70838]

    Consolidate tinycthread bits into libbu

  • Cliff Yapp Cliff Yapp committed [r70837]

    eliminate a couple OS flags to make regress-repository happy

  • Cliff Yapp Cliff Yapp committed [r70836]

    Similar tweaks to adrt embedded tinycthread

  • Cliff Yapp Cliff Yapp committed [r70835]

    Fix Windows pieces of librt embedded tinycthread

  • Cliff Yapp Cliff Yapp committed [r70834]

    mark testing vars as advanced

  • Cliff Yapp Cliff Yapp committed [r70833]

    Turn off the use of CPU_ZERO and MD_CONF_ENABLE_PROGRESS bits that needed mtSignalWaitTimeout. Strip down tinycthread bits to those actually used and build them as part of librt. Do the same for adrt, and remove tinycthread as a separately managed library. This is a bit ugly, but is intended as a stopgap measure until after the next release. At that time, look at moving adrt to C++11 threads and updating the gct_decimation code to newer version, using mmthreads.h. Untested with MSVC.

  • Cliff Yapp Cliff Yapp committed [r70831]

    Globals bite us again - CMAKE_C_STD_FLAG -> C_STANDARD_FLAGS. This time make sure C_STANDARD_FLAGS is non-empty except in the MSVC case, so CMake doesn't silently do the wrong thing.

  • Cliff Yapp Cliff Yapp committed [r70830]

    not right...

  • Cliff Yapp Cliff Yapp committed [r70829]

    get as far as the undefined reference to _tthread_timespec_get error reported earlier

  • Cliff Yapp Cliff Yapp committed [r70828]

    RT_G_DEBUG isn't always defined to what we need/want

  • Cliff Yapp Cliff Yapp committed [r70827]

    Fix MSVC mged build

  • Cliff Yapp Cliff Yapp committed [r70826]

    Various tweaks for MSVC building

  • Cliff Yapp Cliff Yapp committed [r70825]

    Pull in the latest changes from tinycthread

  • Cliff Yapp Cliff Yapp committed [r70822]

    stash some thoughts on how to diff sets of ray partitions.

  • Cliff Yapp Cliff Yapp committed [r70819]

    Don't need to copy to the sh directory in CMAKE_BINARY_DIR - copying to bin is the important part. In particular, the configure_file piece was marking files in the src directory for distcleaning when CMAKE_SOURCE_DIR == CMAKE_BINARY_DIR.

  • Cliff Yapp Cliff Yapp committed [r70816]

    Sigh - even 120 wasn't enough. Turn this off until regress-repository is handled.

  • Cliff Yapp Cliff Yapp committed [r70815]

    60 seconds isn't long enough - regress-repository can exceed it.

  • Cliff Yapp Cliff Yapp committed [r70814]

    let distclean know to remove the scripts

  • Cliff Yapp Cliff Yapp committed [r70811]

    whoops - remove libremrt from test list

  • Cliff Yapp Cliff Yapp committed [r70810]

    if we're bailing, delete df

  • Cliff Yapp Cliff Yapp committed [r70809]

    whoops, forgot to update the README for zlib 1.2.11 - fix

  • Cliff Yapp Cliff Yapp committed [r70808]

    Update libgdiam to version 1.0.3 (thanks to Dr. Har-Peled for the heads up). MIT license is an option now - select that for our use.

  • Cliff Yapp Cliff Yapp committed [r70807]

    Restore (currently) needed CMAKEFILES calls, put check back in distcheck but fix the FORCE_DISTCHECK setting (needs to get the setting from configure_file since distcheck_repo_verify isn't run directly by the user.)

  • Cliff Yapp Cliff Yapp committed [r70796]

    right - new format option, so update regress

  • Cliff Yapp Cliff Yapp committed [r70795]

    slew of nirt diffing related changes.

  • Cliff Yapp Cliff Yapp committed [r70794]

    fixes for linux build

  • Cliff Yapp Cliff Yapp committed [r70791]

    Fall back to the stub case if Threads_FOUND fails, rather than fatal error.

  • Cliff Yapp Cliff Yapp committed [r70790]

    We still need the HAVE_PTHREAD_H test - libbu/parallel.c keys its definition of rt_thread_t off of this test for pthread platforms

  • Cliff Yapp Cliff Yapp committed [r70789]

    Put in the start of diff reporting

  • Cliff Yapp Cliff Yapp committed [r70784]

    if we're going to shoot, need to make sure we're set up for it

  • Cliff Yapp Cliff Yapp committed [r70783]

    whoops - check for diff before trying anything

  • Cliff Yapp Cliff Yapp committed [r70779]

    typo

  • Cliff Yapp Cliff Yapp committed [r70778]

    checkpoint

  • Cliff Yapp Cliff Yapp committed [r70775]

    checkpoint

  • Cliff Yapp Cliff Yapp committed [r70774]

    checkpoint

  • Cliff Yapp Cliff Yapp committed [r70773]

    add a note that this validation step should be pushed up to the fmt command...

  • Cliff Yapp Cliff Yapp committed [r70772]

    Add some guards for report printing to avoid inactive keys getting used

  • Cliff Yapp Cliff Yapp committed [r70771]

    need netinet/in.h for htol declaration

  • Cliff Yapp Cliff Yapp committed [r70757]

    initialize out_old

  • Cliff Yapp Cliff Yapp committed [r70756]

    checkpoint

  • Cliff Yapp Cliff Yapp committed [r70755]

    Start working on a man page specific to NIRT's formatting functionality. This is borderline between a file format and command documentation... we are essentially using it as a file format by defining pre-packaged NIRT formatting files. Also, the topic is sufficiently complex/verbose that it warrants its own discussion and examples rather than hiding it in a bigger NIRT man page.

  • Cliff Yapp Cliff Yapp committed [r70731]

    For diff cataloging, the one-shot nirt output container is inadequate. Start breaking it down into components.

  • Cliff Yapp Cliff Yapp committed [r70730]

    stash a couple reference links

  • Cliff Yapp Cliff Yapp committed [r70729]

    More input logic for diffing

  • Cliff Yapp Cliff Yapp committed [r70728]

    handle comments better

  • Cliff Yapp Cliff Yapp committed [r70727]

    read in hit

1 >