Back
[12:52:31] <terkaa__> Evenning
[15:26:14] <bpuk> evening all
[15:26:41] <bpuk> sign error in the G71 code is fixed, and pushed to my forked repo - now to figure out how to do a pull request
[15:33:55] -linuxcnc-github:#linuxcnc-devel- [13linuxcnc] 15bpuk opened pull request #212: Ben potter/g71 (06BenPotter/G71...06BenPotter/G71) 02
https://github.com/LinuxCNC/linuxcnc/pull/212
[15:36:25] <bpuk> ok. that's wierd - the build failed on downloading a translation module :D
[15:39:30] * jepler clicks "restart build" on travis-ci
[15:39:43] <bpuk> ah-ha - where the heck is that button? :P
[15:40:33] <jepler> I don't think you get it
[15:41:02] <bpuk> oh good - I thought I was going slightly mad there
[15:44:57] <jepler> the new build has gotten down to the tests
[15:55:16] <jepler> tests passed :) but there are no g71 tests as far as I see so :(
[15:56:57] <bpuk> not yet - I wanted to merge where I had got to in 2012 before I made any changes
[15:57:08] <bpuk> since that version has been run on metal a lot
[15:58:44] <bpuk> I also need to look at the unit tests and work out how to make them
[15:59:31] <jepler> look at tests/interp/g81/g17/g98 (for example)
[15:59:56] <jepler> you'd write your own .ngc file, one that you know runs properly on real metal
[16:00:11] <jepler> take a copy of test.sh and modify it to call your ngc file
[16:01:02] <jepler> then you run the test, move "result" to "expected" (since you know the result is right on your real hardware we'll assume it's right in the simulated interpreter environment), and commit the 3 files
[16:01:14] <jepler> to run tests, you . scripts/rip-environment, then runtests path/to/specific/testdir
[16:01:22] <jepler> only in an RIP build
[16:01:32] <jepler> without argument, it runs all tests, not "tests under the current directory".
[16:01:44] <jepler> I think that's mostly what you would need to know to write a test
[16:01:55] <jepler> you'd eventually want to have a series of tests that cover all the interesting cases you could think of
[16:02:39] <bpuk> _all_ of them? :P
[16:03:51] <cradek> bpuk: I'm on the road and can't talk much, but want you to know I'm happy you're working on this again.
[16:04:50] <bpuk> thanks - I gave up due to conflicting requirements back in 2012. The patch we're up to is the one I used from 2012 to 2015 (when I got access to a bigger lathe)
[16:05:14] <bpuk> I'm glad to be back working on it - it's a nice challenge
[16:05:43] <cradek> ooh 3 years of use is a great sign
[16:06:13] <bpuk> it is - but at the same time, I never did 'crazy' things within the loop
[16:06:26] <bpuk> so it's not as thoroughly tested as I'd like
[16:06:41] <cradek> like all code!
[16:12:40] <bpuk> well, there is 'well tested' and then there is 'proved'
[16:12:48] <bpuk> I'll settle for the first ;)
[16:14:02] <cradek> thinking about how to break it (knowing the code) and not being able to is a very big plus
[16:14:37] <bpuk> oh, I haven't tried too hard to break it yet, that's coming up next
[16:26:25] <bpuk> Hmm... I haven't tested G5, G5.1, G5.2, G5.3 yet
[16:29:32] <jepler> I think those aren't supported during cutter compensation, I would be fine saying they are unsupported in G71 too
[16:29:45] <jepler> afk
[16:48:24] <bpuk> Terrifying comments... // TESTME!!! MORE THOROUGHLY !!!KL
[17:07:38] <bpuk> Question: Should G71 respect CSS?
[17:56:27] <andypugh> Where does uspace send the output that would normally go to dmesg?
[18:03:33] <pcw_mesa> console
[19:17:05] <cradek> bpuk: oh yeah, definitely
[21:30:07] <skunkworks> dragon?