* read timing factors for string transmission by external
program using new xmlrpc query
- main.get_tx_timing
respone - #samples: samplerate: transmit_time (secs)
* read char rate table using external xmlrpc query
- main.get_char_rates
table structure
char, ascii-rep', #samples, samplerate, transmit_time (secs)
* added hidden macro <CPS_TEST> which simulates transmission
of test string and measures the character per second rate
for the modem in use.
- <CPS_TEST> uses internal "Jabberwocky" string
- <CPS_FILE:[full pathname to file]> uses external text file
- <CPS_STRING:ascii text string...>
- produces report similar to:
CPS test
text: /home/dave/dev.git/all.git/pskr-tests/paris.txt
mode: PSK-500R
# chars: 2400
xmt time: 56.580000
data time: 52.480000
data samples: 419840
sample rate: 8000
chars/sec: 45.731707
~WPM: 457.317073
* Allow writing wav file during timing test
* High Speed WAV
- Add macros to enable writing WAV file in at maximum
character throughput.
- disables codec output
- use this for creating wav test files
* Suppress waterfall display for CPS and WAV tests
* Remove PSK printf debug statements
* Fix MFSK xmt 0x05 --> 0x04 during CPS testing
* Corrected macro editor hints for <SKED:...
* Time of execution is specified as HHMMSS
where SS is optional and will be substituted as
"00" if omitted from the macro text. Time is
is Zulu.
* Date of exection is specified as YYYYMMDD
where YYYY is year, MM month, and DD day.
* Mod to correctly save state of word wrap selection
- Tx and Rx widgets can be set for word wrapping
- Mod insures that the last selection for each control
is restored the next time fldigi is executed.
* Modified timeouts and timeout logging
- newer modems timed out waiting for
release from Tx to Rx state.
* Changed sending 0x06 character to ARQ client
from end of ARQ tx buffer event to actual end
of transmit event.
* Some methods in waterfall are being called
before a modem has been instantiated. Possibly
due to threads being assigned to separate cpu
processors. This cannot happen on h/w with a
single cpu thread implementation.
* removed source files for Cfft, the original fast fourier
transform used in fldigi. Cfft has been replaced by
g_fft, the fast fourier transform C++ template based on
public domain code by
John Green <green_jt@vsdec.npt.nuwc.navy.mil>
* Changed fftfilt implementation to use g_fft class vice
Cfft class.
* Removed multi-channel decoding when not visible or configured
for decoding when not visible.
* Added macro tag for sending image. Similar to <File:xxx>
<IMAGE:[full pathname to image file]>
A file finder will open for image file selection when the
macro tag is selected from the pick list.
* re-scaled MFSK soft-bits: improved very-weak signal decode
* added CWI detection and avoidance
* removed "Hide", "Save", and "Abort" buttons from receive
pic dialog. Dialog remains open after changing modes.
Close using the system "X" close button.
* Added LOG_DEBUG statement to indicate when an
rsid code is not assigned to a modem type.
* Added code to disallow
<!MODEM:SSB> and <!MODEM:ANALYSIS>
macro tags.
* Added code to clear macro queue when T/R returns
to receive using the T/R button.
* Added assigned(trx_mode m) method to class cRsId
- used for testing whether Tx loop should invoke
the cRsID::send(...) method
* Corrected position of dtmf->send() in tx loop
* These changes are an attempt to catch and/or document start
up failures on Windows systems.
* Standardized the fatal exit process and message handling
* Changed system calls in util function create_directory(...)
* Changed name of DATA.files to "data" and placed directory in
the home directory for specific target OS.
- Added creation of data directory to main()
- This effects the location of Synop data files!
* Changes needed to improve double RsID detection
* Removed resample test and dependent code
* Reimplemented FFT bucket array
- removed array pointer arithmetic
- memory copy and initializers used to update
the bucket array
- modified search routine to insure that all elements
of FFT bucket array are tested.
* Deprecated use of hash tables for initial RsID search
- search sensitivity and hash table use were incompatible
David Freese