> Also, is this a Debian bug? > I am not able to reproduce this on my Debian machines.
I ran into this problem when trying to compile https://github.com/analogdevicesinc/scopy/. Specifically the first compile failure occurs when trying to compile https://github.com/analogdevicesinc/scopy/blob/master/src/HistogramDisplayPlot.cc. I did try to reproduce with simple examples (since scopy is complex to build) but like you was unable to do so. At the same time I was unable to see anything wrong with scopy's code. Using the patch linked above fixed the compile failure and I was able to build scopy successfully. So I can also confirm the patch applies cleanly. > Is this really a libvolk2-dev bug? The upstream discussion leads > to the conclusion that this is a now-fixed compiler bug. The issue dicussion does indeed seem to conclude that this is a compiler bug in gcc <= 6. However I do see the problem even with gcc 10 that is part of Debian/Bullseye. I linked to the issue above precisely since I was a bit surprised. Despite the discussion in that issue there is an open pull request at https://github.com/gnuradio/volk/pull/455 from which the patch linked above was extracted. The patch only introduces "_cplusplus" header guards that does seem like the right thing to do since libvolk intends to support compilation with both C and C++. Given the nature of this patch, regressions are very unlikely. A quick check shows at present only Gnuradio plugins and Gqrx depend on libvolk. Most of the former are implemented in C. Perhaps this may be the reason why the bug (?) was not exposed as yet. If you are convinced to let this patch through I will be happy to help in any way I can.
