Baud rate issues (major issue seems to be with upstream)

[m1ari]

Mostly to note them:

1. 600 / 1200 baud missing in the interface.

2. Higher baudrates don't seem to decode. This appears to be an issue in upstream.
   This is testing on windows with dl-fldigi 3.1 and fldigi 3.21.77
   100 baud works on both
   150 baud only worked on dl-fldigi
   200 baud only worked on dl-fldigi
   300 baud only worked on dl-fldigi

A couple of screenshots at https://www.dropbox.com/sh/d4xrs4cpd6w49h0/mzg3K2yjFh

I also tested 100bd and 300bd with the new linux dl-fldigi with results matching the above.

Additional testing has been done with dl-fldigi (non hab mode) and fldigi transmitting to each other via a virtual audio cable at 200bd. dl-fldigi could decode fldigi but fldigi couldn't decode dl-fldigi.

I've also tested fldigi talking to fldigi in a similar setup which works at 100 baud but fails at 150 (further confirmation of it being an upstream issue)

[m1ari]

https://lists.berlios.de/pipermail/fldigi-devel/2013-July/001054.html
This suggests it was working in fldigi commit 645acab (29/01/2013), but broke sometime between then and 27/1/2013 (date on that posting).
The release prior to that was 3.21.65 so assumption is it broke sometime since then (I'll try to narrow down when later on)

[jamescoxon]

seems that filter code added between 3.21.72 and 3.21.73 has created the problem, filter code doesn't suit the higher baud rates. Code added to only use this filter if the baud rate is 100 or less. Seems to be working: 501f869

[hexameron]

ROFLMAO: the filter is calculating the sine of integer multiples of pi/2

looks like it should be using the scaled value x, not the index i

[hexameron]

Looking back through the git history I see that Stefan was trying to divide by the Sqrt of 2, so should have used (i_pi/4) instead of (i_pi/2) - This still looks odd but removes the need for "f *=1.4", added later by experiment.
jamescoxon@dfc2bd8