# Playing with 3.7

**URL:** <https://www.ruby-forum.com/t/playing-with-3-7/240179>\
**Category:** GNU Radio\
**Created:** [April 1, 2014, 3:16am UTC](https://www.ruby-forum.com/t/playing-with-3-7/240179 "2014-04-01T03:16:12Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![musicdenotation](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/m/ea666f/32.png) [@musicdenotation](https://www.ruby-forum.com/u/musicdenotation)\
**Post date:** [April 1, 2014, 3:16am UTC](https://www.ruby-forum.com/t/playing-with-3-7/240179/1 "2014-04-01T03:16:12Z")

</div>

So, I have this flow-graph that mostly works under 3.7 now, except for  
something that’s horrible.

In my flow-graph, I have two layers of selectors, to select the input  
for a WX GUI FFT sink. The layers select either the “raw” or “filtered”  
chunk of bandwidth (coming off the 4 hardware sources), and the other  
layer selects which source. This worked like a champ in 3.6.5.1–never  
a problem.

Under 3.7.4 (basically, today’s master), when I go to select which  
input, there’s a significant _pause_, and then the change is applied  
after up to  
about 2 seconds of “hang”. If I go to select the “raw” vs  
“filtered”, the whole application freezes solid–the GUI freezes up, no  
displays are updated,  
and the hardware sources start getting massive overruns.

Now, when this thing is running, it consumes about 8% of my system, and  
when it “freezes” it consumes apparently much, much less.

Now, I know that “selectors” are actually kind of a hack that stop the  
flow-graph, reconfigure it, and start it up again. So, maybe there’s  
some kind of deadly-embrace going on.

I can reimplement using adders and multipliers if I have to, but gosh,  
I’d rather not have to. I’m going to try to reproduce with the  
smallest flowgraph that shows the problem, but a simple quick test  
tonight worked flawlessly. Sigh.