Yeah, I hear you, issue for us is we are on el6 for at least a couple more years as it's what our primary customer uses so all our other tools are built to it. Sofar I ripped deps from the windows build and have compiled most of them along with current GCC. Making decent progress(working a couple hours a week on it), when I'm done I should be able to post said big nasty bash script. However I didn't think of a chrooted modern distro, I know there was some docker work going on. That shouldn't take any real performance penalty so may be a way to go if a chrooted Ubuntu doesn't crap out with a 2.6 kernel.
On May 18, 2016 9:26 AM, "Andy Walls" <
address@hidden> wrote:
On Wed, 2016-05-18 at 03:11 -0400, Anon Lister wrote:
> Andy,
[snip]
> Since I'm replying in the pybombs thread, I did try it and it broke
> pretty badly. Missing deps and such, I'd be willing to try to help get
> with that after I get the build down by hand and know what I can use
> from yum and what I have to source build (which I expect is almost
> everything).
Building on SciLinux 6.x (and probably CentOS 6.x) with PyBOMBS is a
fool's errand. Take it from a fool who tried.
PyBOMBS should *not* target RHEL/CentOS/SciLinux 6.x as the Python is
too old, and the old Qt4 version forces maximum versions for Qwt, SIP,
PyQwt, and maybe other libraries. M4, autoconf and automake are too
old. libcurses.so is also broken in a funny way, so libuhd examples
won't compile without fixing it. A few items need patches to build. It
goes on and on....
A big nasty Bash script, a chroot-ed modern distro install, or a more
capable build system (like bitbake), is needed for GNURadio on those
distros.
> Did get el7 build without much issue, don't think I needed to source
> build GCC or anything, there were only a few packages that needed
> source build, boost I think was the only big one.
And since el7 *is* available to people, any hard work on el6 is
something I would not encourage.
Regards,
Andy