WSJT-X/lib/extract4.f90
Bill Somerville b048669c02 Merged from trunk:
------------------------------------------------------------------------
r8060 | k1jt | 2017-09-01 13:51:42 +0100 (Fri, 01 Sep 2017) | 2 lines

Add a link to G3WDG doc on using QRA64 for microwave EME.

------------------------------------------------------------------------
r8061 | k1jt | 2017-09-01 17:22:19 +0100 (Fri, 01 Sep 2017) | 1 line

Fix a misspelled word.
------------------------------------------------------------------------
r8062 | bsomervi | 2017-09-01 21:10:35 +0100 (Fri, 01 Sep 2017) | 7 lines

Rationalize NA contest mode

Generic message packing and unpacking routines now understand antipode
grid contest messages.  These messages  are now recognized as standard
messages in  message response processing and  dealt with appropriately
when contest mode is selected and applicable (currently FT8 and MSK144
only).
------------------------------------------------------------------------
r8063 | bsomervi | 2017-09-01 21:43:45 +0100 (Fri, 01 Sep 2017) | 1 line

Fix issue compiling with Qt older than v5.7
------------------------------------------------------------------------
r8064 | bsomervi | 2017-09-01 22:29:02 +0100 (Fri, 01 Sep 2017) | 7 lines

Fix issues with type 2 compound calls in contest mode

Message generation in  contest mode now generates the  correct Tx3 for
type 2 calls.

"<type-2> 73"  is a free  text so  needed special handling  in message
processing.
------------------------------------------------------------------------
r8065 | bsomervi | 2017-09-01 23:22:20 +0100 (Fri, 01 Sep 2017) | 11 lines

Improved message generation for type 2 calls in contest mode

These attempt  to ensure that  a prefix is  logged by the  QSO partner
even if the compound call holder user Tx3 to tail-end a QSO.

The  type  2 message  generation  options  are largely  overridden  in
contest mode as only a few options make sense. Key is that Tx1 may use
only the base call when calling split is necessary, this requires that
both Tx3 and Tx4 have the full compound call otherwise the QSO partner
will never see the  full call until it is possibly  too late i.e. post
logging.
------------------------------------------------------------------------
r8066 | bsomervi | 2017-09-02 00:28:44 +0100 (Sat, 02 Sep 2017) | 5 lines

Fix erroneous auto stop critera for auto reply in FT8

We cannot assume that a "DE <dx-call>  <anything>" is or is not for us
so  we must  continue calling  and  risk possible  QRM. Calling  split
avoids this issue.
------------------------------------------------------------------------



git-svn-id: svn+ssh://svn.code.sf.net/p/wsjt/wsjt/branches/wsjtx-1.8@8067 ab8295b8-cf94-4d9e-aec4-7959e3be5d79
2017-09-01 23:55:56 +00:00

70 lines
1.6 KiB
Fortran

subroutine extract4(sym0,ncount,decoded)
use packjt
real sym0(207)
real sym(207)
character decoded*22
character*72 c72
integer*1 symbol(207)
integer*1 data1(13) !Decoded data (8-bit bytes)
integer data4a(9) !Decoded data (8-bit bytes)
integer data4(12) !Decoded data (6-bit bytes)
integer mettab(-128:127,0:1) !Metric table
logical first
data first/.true./
save first,mettab,ndelta
if(first) then
call getmet4(mettab,ndelta)
first=.false.
endif
!### Optimize these params: ...
amp=30.0
limit=10000
ave0=sum(sym0)/207.0
sym=sym0-ave0
sq=dot_product(sym,sym)
rms0=sqrt(sq/206.0)
sym=sym/rms0
do j=1,207
n=nint(amp*sym(j))
if(n.lt.-127) n=-127
if(n.gt.127) n=127
symbol(j)=n
enddo
nbits=72
ncycles=0
ncount=-1
decoded=' '
call interleave4(symbol(2),-1) !Remove the interleaving
call fano232(symbol(2),nbits+31,mettab,ndelta,limit,data1, &
ncycles,metric,ncount)
nlim=ncycles/(nbits+31)
!### Make usage here like that in jt9fano...
if(ncount.ge.0) then
do i=1,9
i4=data1(i)
if(i4.lt.0) i4=i4+256
data4a(i)=i4
enddo
write(c72,1100) (data4a(i),i=1,9)
1100 format(9b8.8)
read(c72,1102) data4
1102 format(12b6)
call unpackmsg(data4,decoded,.false.,' ')
if(decoded(1:6).eq.'000AAA') then
! decoded='***WRONG MODE?***'
decoded=' '
ncount=-1
endif
endif
return
end subroutine extract4