Marcus got it: there are old headers around in some system prefix (possibly including "${CMAKE_INSTALL_PREFIX}/include"), which are being picked up by CMake and included ("-I...") -before- the internal-to-GR-build ones. The GR API changed, and because the old header is being used with the new code, the error is generated. I've investigated this issue & could figure out no good fix because of the way CMake handles the "include directories" internally between targets and the current build. Hence, my recommendation is to not build a new GR while an old one is installed in system prefix (etc). - MLD
On Thu, Jan 2, 2020 at 11:16 AM Marcus Müller <marcus.muel...@ettus.com> wrote: > Hi Tom, > > funky distro! > That API used to be string (of a serialized PMT) and now is PMT > directly. That happened somewhere around July. > Thus, this slightly looks like you still have some older header files > lying around? Or some outdated SWIG output that somehow ends up in your > build? > Could you try in a clean container? I'm trying here locally, but > honestly, I've never used slackware before, and thus I need to learn > every packager tool in the process, and that's kind of a burden. > > Best regards, > Marcus > On Thu, 2020-01-02 at 15:18 +0000, Tom Crane wrote: > > I am attempting to build the v.3.8 source under Slackware64 Linux > > and > > can't get past this and a few subsequent errors. > > > > Building from the git source results in the same failure. > > > > Build system details: > > o gcc 9.2.0 > > o SWIG version 4.0.1 > > o Boost 1.72.0 > > > > My build scripts and build logs are here > > https://www.mklab.rhul.ac.uk/~tom/tvdx/gr/build_problems/ > > > > Please advise. > > > > Many thanks > > Tom Crane > > > > > -- Michael Dickens Ettus Research Technical Support Email: supp...@ettus.com Web: https://ettus.com/