introduce ipcbuf_relock_write to allow a writer to reconnect in WCHANGE without relinquishing the writer lock
update changelog, fix other minor syntax issues
fix incorrect extern cpp for cuda_runtime
ensure mark_filled on header block writes correctly set EOD marker, such that the header block xfers are aligned with data block xfers. Fix spelling, unify linting, syntax and copyright information across library and applications.
remove debug values from dada_dbmergedb input_thread method
Fixed last read the position since the client read was not recorded
revert unneeded change to data_hdu_unlock_read
minor correction to Management/psrdada.pc.in
fix bug when a writer closes ring buffer without writing any data
check if data_block is open before closing it in dada_hdu_unlock_read
add dada_dbwait app that returns when the specified ring buffer is created
Allow both viewer and viewing to use ipcbuf_get_next_readable
add support for interrupting ipcbuf_get_next_readable
fix bugs with dada_header and ipcbuf_get_next_readable, moving this method to require a viewer connection
Fix the delay function to always be zero
Fix the delay function to always be zero
install scripts and pkgconfig when building with cmake
files written now have user write permission and other read permission
prevent segfault when attempting to delete non existent ring buffer
ipcbuf_destroy now also detaches from SHM when deleting it
add support for multiple connnection attemps to ring buffers
remove trailing whitespace and fix compilation warnings
ensure dada_diskdb calls dada_hdu_unlock_write on exit
ensure CUDA headers are installed with CMake
adjustments to support building of aarch64 and fix compiler warnings
Set the sync->r_states and sync->w_state attributes
Remove unused dependence on emmintrin.h
Fix incorrect pointer order for dada_client_read for GPU ring buffers
Merge branch 'instrument-refactor' of ssh://git.code.sf.net/p/psrdada/code into instrument-refactor
ignore files with names ending in tilde
compiles with Apple clang version 13.1.6
fix compilation warnings under gcc-10
improve error checking on cudaSetDevice
support longer key/value strings in ascii header
correct verbose logging in dada_dbgpu
ensure variables are initialised
Fix incorrect AC_CONFIG_SRCDIR
Return an error if ipcio_close cannot close an open block
Add missing CUDA cflags for apps
removing old 3rdparty code from .gitignore
ensure Cmake builds correct library version, convert csh scripts to sh
Refactor PSRDADA instruments out of the psrdada-code repository here
Add -lm depdendency
Adjust compiler options to use AM_CPPFLAGS
dropped cmake minimum requirement to 3.13.4
Improved cmake build and added pkg-config, psrdada_cflags, psrdada_ldflags
remove references to 3rdparty
shift 3rd party timing code into the main library
Adding basic tests for the library, fixing bugs in command_parse
Add a pkg-config file
Refactor all instruments, perl scripts, php web pages out of this repository, shifting them to psrdada-instruments
Correcting tests in autoconf
add option to skip memcpy on write
fix libtool search path
Added support for binding to specific IB interface
updated aspects to web system by others
Added a fast compute for converting equatorial to galactic
Support for non 4096 byte headers in various places
added some more verbose output, check cal_app_pos_iau return value
switch to use shfl__sync variants for future CUDA compatibility
Added now tied-beam plotted
start on a 1s tick
Added support for Search mode DM parameter to be passed
check sofa return values
Improved performance of dbcalib
misc MOPSR backend updates
Ensure an undef response is returned as an empty string
Added support for fractional beam position furbies
dropped number of preferred modules to 11
Added support for fractional beam position furbies
Added DP4A implementation to HIRES beam former
added support for multi-beam furbies
add check for omp_get_num_threads
ignore for mopsr_eq2gal
Use DB to chose pulsars for residuals page, allow ta
Produce S/N, gof, and use the former to filter ToAs
residuals.php: format the link differently for better workflow
Improved coordinates / VOEvent
Include checksum in transient emails
Provide a 2k 25s proc file
Store last UTC on disk
Fit for F0/1 before plotting vs postfit
Transfer all_candidates to HPC
Link on Asteria to 1644 snrs when clicked on the image
Display P0, DM, and cadence on results page
asteria workaround for smooth_mmw failure
Fix a problem with printing P0 DM
Clean up a bit, update paths on ozstar
Merge branch 'master' of ssh://git.code.sf.net/p/psrdada/code
Merge branch 'master' of ssh://git.code.sf.net/p/psrdada/code
do not use OPENMP for ring buffer disconnect or CUDA zeroing
do not try to bind memory when no NUMA node specified
Use OPENMP_CFLAGS for the link phase
ensure the required openmp flags are available for libraries compiling against psrdada
Added RDMACM_LDFLAGS for psrdada_ldflags
added extern C protection, removed deprecated header
ipcbuf_get_[nfull|nclear|sodack|eodack] now correctly return the max value for multi reader ring buffers
change the interpreter to use any python
dont let -fopenmp be passed to nvcc
use openmp for faster DB creation