"BruteFIR is a software convolution engine, a program for applying long FIR filters to multi-channel digital audio, either offline or in realtime." It's extremely useful for very high performance audio filtering, for example applying complex filters to multichannel audio. It integrates with Jack, and this is probably the easiest way to hook it into the audio chain. In particular it's very useful for applying filters generated by DRC that perform "Digital Room Correction" (see http://drc.wildgooses.com ) It's a pretty trivial ebuild install, and I would be happy to maintain this going forward. Reproducible: Always Steps to Reproduce:
Created attachment 36261 [details] brutefir 1.0 ebuild
Hi Ed, thanks for your ebuild. I had a brief look at it and found some minor issues. There is no intent to scare you off, but I'd appreciate if you'd handle these. - your list of keywords lists a whole list of architectures; did you really test it on all of those? - this is a new ebuild, therefore arches should not be marked stable - dependency to dev-libs/fftw version 3 or higher seems to be missing (possibly more dependencies missing?) - the src_unpack() you provided can be left out since the default will do fine here - the documentation (CHANGES, LICENSE, README etc.) does not get installed - can't you use the einstall instead of the make install? - upon starting brutefir from commandline it complains about a missing configuration file. is it possible to provide a default configuration file to make it work out of the box? - could you include a src_test() to run the build-in test-case? - version 1.0a is out now :-)
Created attachment 37531 [details] Brutefir 1.0a ebuild
Hi Frank. Thanks for your very helpful comments - that kind of feedback is very helpful to teach muppets like me to do better! I have attended to most of the things that you raised in this new ebuild. However, there are a few things that I would like to check with you: I included the README, CHANGELOG and LICENCE, but I feel that these are probably redundant. The README in particular just refers the user to the web page for the documentation. However, they are in there. What I have done is add an "einfo" line to refer the user to the website for the documentation as well - I think this is probably the most useful thing I have copied the example config files to /usr/share/brutefir which gives people something to look at, however, to be honest, to do something useful with this software you need to be capable of building your own FIR digital filters and reading the instructions, so having something working out of the box is not too essential. This isn't really an end usertool, but more of a massive filtering engine that may be useful either to an enthusiast, or as part of another program. I couldn't see the test case that you referred to, so that hasn't been added? Can you please point me to that if I missed something? Thanks for the feedback - I think this should now be satisfactory. I am happy to continue to maintain this for new updates of course.
Created attachment 37532 [details] Brutefir 1.0a ebuild Nuts. I previously uploaded an old version which didn't have the /usr/share stuff in it. This one is complete.
Created attachment 37533 [details] Brutefir 1.0a ebuild Sorry to waste space. I think this cleans up a few things like the RDEPEND, and not needing flag-o-matic
Hi Ed, your latest ebuild looks good. I cleaned a few minor things, added the ~amd64 keyword and committed it to the portage cvs tree. Thanks for your work!
Created attachment 39410 [details] Updated ebuild to honor portage CFLAGS The default Makefile does not use the CFLAGS variable, so I added a little sed to the src_compile() to change the Makefile slightly. In a DSP package like Brutefir having the right flags does actually make a difference! No other bugs found so far, recommend this be marked stable on x86.