GNU Radio – the Free and Open Software Radio Ecosystem
GPL-3.0 License
Bot releases are visible (Hide)
Published by marcusmueller about 5 years ago
This is the second release candidate for GNU Radio 3.8.
gr_modtool
: Paths with spaces in them used to breakPublished by marcusmueller over 5 years ago
v3.8.0.0-rc1 Changelog
All notable changes to GNU Radio will be documented in this file.
The format is based on Keep a
Changelog and this project adheres
to Semantic Versioning, starting
with version 3.7.12.0.
Older Logs can be found in docs/RELEASE-NOTES-*
.
This is the release candidate for GNU Radio 3.8.
For brevity reasons, this document does NOT contain a complete CHANGELOG
comparing 3.8.0.0 to 3.7.0, because the development model changed
underway, and there's no human-feasible way to detangle the changes from
more than six years of development.
The major user-relevant changes will be noted here in the actual
release.
What we can do here is point out that we've got a glorious amount of 215
individual contributors that added at least one commit since 3.7.0.
This is a big step towards the future of the world's most popular
FOSS SDR framework.
Let us ride eternal, shiny and chrome.
Published by marcusmueller over 5 years ago
udp_source_sink
flakinesspolar_encoder
/_common
: Memory leakrpcbasic_inserter
Published by marcusmueller over 6 years ago
fmdet_cf
's derivative coefficients were wrong.skiphead
used to incorrectly handle tags, now properly shiftsqa_packet_format
: Unit test used to potentially lock up due to incorrect conditionalsclock_recovery_cc
, crc32_bb
: Accessing the [0]
element of an empty vector is UB, even if you don't use the result afterwards.polar_encoder
/_common
: Fixed multiple out-of-bound accesses due to insufficient vector length checksfft_filter
, filter_delay
: Accessing the [0]
element of an empty vector is UB, even if you don't use the result afterwards.Published by marcusmueller over 6 years ago
Published by marcusmueller over 6 years ago
boost::posix_time::[milli]second
takes integer numbers, and we now at least cast the floats to integers, as Boost 1.67 forced us to realize we're (incorrectly) using floating point herfloat_array_to_int
: int32 limits were wrongly hardcoded.float_to_complex
when accessing the second input_items element in the single-input casePublished by marcusmueller over 6 years ago
unique_ptr
)log
vs logf
performance bugspmt::mp("string")
calls (instead of cached PMTs) performance bugrandom.h
: Uniform integer Generation depended on Boost version > min versionrandom.h
: time-dependency when seeding with 0
replaced with actual standard seedget_tags_in_range
used v.resize(0)
to clear vectors, which lead to unnecessary memory reallocationslogger_ptr
now always a pointer, no matter whether logging is enabled; fixes heap corruption bugbuffer[_reader]_sptr
superfluous grab/release performance bugequal()
comparisongr_modtool
makexml processing, leading to improved reproducibility of buildsfmdet_cf
now has GRC filefastnoise_source
: non-threadsafe libc
rand()
replaced by XOROSHIRO128+simple_copy
Ctrlport exampleclose()
multiply_matrix
formerly used "magic int" value where its API expected enum (and that int wasn't covered)tags_strobe
: superfluous tags additionvector_sink
: multithreading hazards on data
accesstuntap_pdu
: MTU-sized IP packet receptionsocket_pdu
: Garbage Collect closed TCP connectionsselective_fading_model
: proper scaling of path lengthadditive_scrambler_bb
: tag triggered resetstream_to_tagged_stream
: tag offset after unlock()
blks2/packet.py
: Flushing for small packetsinterpolating_resampler
: GCC 4.0 compile errorconstellation
: 16QAM demapper was wrong for one constellation pointchunks_to_symbols
: mutex-guard LUT against modification during useand
-> &&
etccrc32_bb
: handle unpacked bytes correctlyofdm_carrier_allocator
: catching wrong allocations instead of crashingchannel_construction.py
: return int frozen bit outputGSL_LDFLAGS
propagationpolar_decoder
used to be flakywindow.cc
: removed NaN due to FP mathlogpwrfft.py
reference scaling integer division buguhd_fft
scalar gain-capabilitygr.WORK_CALLED_PRODUCE
and gr.DONE
gr_modtool
: Autocompletion, line editinggr_modtool
: copyrightholder
optiontag_share
block: take tags from one two input streams to one output streamtag_gate
: tag filteringfile_source
: configurable file_begin
stream tagmoving_average
: Vector mode (with element-wise MA)plateau_detector
: threshold get/setcostas_loop_cc
: Optional loop variable outputscorrelate_access_code_tag
blockglfsr
exposed through Pythoncc_encoder
symbol_sync_{cc|ff}
uhd_app.py
: selectable time/clock sourcesrecv()
call's timeout parameterizablegr_??int
typesgr_modtool
: Block template now uses numpy.float32
instead of float. 32bit float is actually the default GNU Radio type.e.vector_sink
can preallocate memory for desired amount of itemsheader_format_default
now sps>1-capablecorrelate_acces_code
uses GR_LOG
uhd_fft
: respect lo_offset
in set_bandwidth
usrp_source
: freq tag now actually reflects last set frequency, even on multiple channelsgr::sys_pri