Reply 11/09/18
Further to the foregoing information.
I have now extensively tested the various compilation alternatives under UBUNTU 18.04.1 with the amd64 RT:Pre-empt kernels indicated previously; the 4.13.13-rt5 (before the latest stable) is the faster.
LinuxCNC 2.80 builds without error in this order: ./configure uspace noauto, dpkg-checkbuilddeps, ./autogen.sh, ./configure --with-realtime=uspace, make clear, make -j4 | tee -a LinuxCNCMake.log [Note: this command may be used to send the terminal output to the *.log file preferred, LinuxCNCMake.log in this case], sudo make setuid, . ./scripts/rip-environment (the last from within the linuxcnc-dev (or top local linuxcnc directory) directory).
Satisfaction of any build dependencies will be forced by the above sequence; they will depend upon your local machine setup.
Following this, in order to have pyvcp and HAL-widget functionality in creating custom control layouts, etc. it will be necessary to install GTK2-compatible versions of both glade and glade-gtk2; this is easily possible via standard trusted repositories.
However, once the above satisfying steps have been completed, and the linuxcnc install appears to function perfectly, in every permutation tried, either using Jared's source or git-hub repository, the outcome of the runtests suite is the same, ie:
*** /home/administrator/linuxcnc-dev/tests/halcompile/extralib: FAIL: test run exited with 2
--- count-and-extra-args.expected 2018-09-07 03:03:14.000000000 +0930
+++ count-and-extra-args.result 2018-09-09 15:42:53.814071238 +0930
@@ -1,26 +0,0 @@
-count=2
-argc=2
-argv[0]=userspace_count_names
-argv[1]=zip
-argv[2]=(null)
-userspace-count-names.0.out
-userspace-count-names.1.out
-
-count=2
-argc=3
-argv[0]=userspace_count_names
-argv[1]=zip
-argv[2]=zap
-argv[3]=(null)
-userspace-count-names.0.out
-userspace-count-names.1.out
-
-count=2
-argc=3
-argv[0]=userspace_count_names
-argv[1]=zap
-argv[2]=zoot
-argv[3]=(null)
-userspace-count-names.0.out
-userspace-count-names.1.out
-
--- count=3.expected 2018-09-07 03:03:14.000000000 +0930
+++ count=3.result 2018-09-09 15:42:54.234110248 +0930
@@ -1,8 +0,0 @@
-count=3
-argc=1
-argv[0]=userspace_count_names
-argv[1]=(null)
-userspace-count-names.0.out
-userspace-count-names.1.out
-userspace-count-names.2.out
-
--- names-and-extra-args.expected 2018-09-07 03:03:14.000000000 +0930
+++ names-and-extra-args.result 2018-09-09 15:42:54.626146656 +0930
@@ -1,33 +0,0 @@
-count=0
-names[0]=swim
-names[1]=swish
-argc=3
-argv[0]=userspace_count_names
-argv[1]=slap
-argv[2]=slop
-argv[3]=(null)
-swim.out
-swish.out
-
-count=0
-names[0]=swat
-names[1]=swap
-argc=4
-argv[0]=userspace_count_names
-argv[1]=slam
-argv[2]=slalom
-argv[3]=slap
-argv[4]=(null)
-swap.out
-swat.out
-
-count=0
-names[0]=sock
-names[1]=slack
-argc=2
-argv[0]=userspace_count_names
-argv[1]=slip
-argv[2]=(null)
-slack.out
-sock.out
-
--- names.expected 2018-09-07 03:03:14.000000000 +0930
+++ names.result 2018-09-09 15:42:55.046185664 +0930
@@ -1,9 +0,0 @@
-count=0
-names[0]=bibble
-names[1]=bobble
-argc=1
-argv[0]=userspace_count_names
-argv[1]=(null)
-bibble.out
-bobble.out
-
--- no-args.expected 2018-09-07 03:03:14.000000000 +0930
+++ no-args.result 2018-09-09 15:42:55.442222444 +0930
@@ -1,6 +0,0 @@
-count=0
-argc=1
-argv[0]=userspace_count_names
-argv[1]=(null)
-userspace-count-names.0.out
-
*** /home/administrator/linuxcnc-dev/tests/halcompile/userspace-count-names: FAIL: checkresult exited with 1
*** /home/administrator/linuxcnc-dev/tests/halcompile/userspace: FAIL: test run exited with 2
loadrt found the test component, and it failed to load
Runtest: 224 tests run, 221 successful, 3 failed + 0 expected
Failed:
/home/administrator/linuxcnc-dev/tests/halcompile/extralib
/home/administrator/linuxcnc-dev/tests/halcompile/userspace-count-names
/home/administrator/linuxcnc-dev/tests/halcompile/userspace
The failure of these tests may not be significant; everything seems to work, particularly after the glade/glade-GTK2 dependencies are satisfied.
It would be wonderful if one of the senior developers close to the code could indicate the likely cause(s), and their significance; at this stage, a faulty data structure, and/or the parsing of that data due to incorrect pointers seems most likely.
Additionally, I have installed 2.80 directly from Jared's repository (ie as he recommends via sudo apt-get *) after correcting the missing dependencies identified during that process; the outcome of the runtests suite is the same.
I have not found any references to the outcome of the post-installation runtests (these or any) or their relationship to any as yet undetermined defects.
Regards
