If you need this for STRF then I suggest to stream the raw data over a UDP port and use pre-obs scripts with netcat to feed it into rffft. That’s really the only way to get the timestamping accurate, as the timestamp of a recorded IQ file may differ by seconds compared to the scheduled start time.
I’ve been doing this for a few years, see this (brief) description: Observation 1037954: Raw data Needed - #12 by cgbsat
This doesn’t help you solve your problem, and I’m not sure why the reinstall wouldn’t work. In the end, you can have a look at the difference in the python files that gnuradio-companion produced. For my python files, the changes are that a blocks_udp_sink block is added, which is connected to the soapy_source, with lines like:
self.blocks_udp_sink_0_1 = blocks.udp_sink(gr.sizeof_gr_complex*1, udp_dump_host, (int(10 * rigctl_port)), 1472, True)
and
self.connect((self.soapy_source_0_0, 0), (self.blocks_udp_sink_0_1))
Note the int(10 * rigctl_port) definition for the UDP output stream.