#linuxcnc-devel | Logs for 2014-08-01

[07:54:45] <cradek> wib
[07:54:56] <cradek> wib!
[07:57:14] <skunkworks__> wib?
[07:58:31] <skunkworks__> only thing I find is references to spooning..
[07:59:39] <cradek> it's me failing to do "/window balance" which is typed like /wi[tab]b[tab]
[07:59:52] <cradek> I guess I missed the /
[08:00:07] <cradek> (laptop keyboard)
[08:01:32] <skunkworks__> heh
[08:02:36] <cradek> an interesting question is why I'm here and not on my way to work
[08:03:06] <archivist> lazing the time away at home is far better
[08:23:34] <pcw_home> hmm stepconf appears to be broken in 2.6 (I just verified though with 2.6pre5)
[08:23:36] <pcw_home> http://www.linuxcnc.org/index.php/english/forum/16-stepconf-wizard/28169-impossible-to-set-good-value-260
[08:28:34] <pcw_home> may be something weird about our environments that stepconf doesnt like
[08:54:15] <skunkworks__> been running the splash screen for 5 days. over and over and over and over at 3000% FO
[08:54:46] <pcw_home> you running wheezy?
[08:57:37] <skunkworks__> yes
[08:58:02] <pcw_home> Hmm does stepconf work for you?
[08:58:16] <skunkworks__> it is the config you had sent me a long time ago.. (I did add the 5i25 so it would load also)
[08:58:51] <skunkworks__> what is the issue? (I have not really used stepconf...
[09:00:06] <pcw_home> in the forum (you cant set step time/space/direction hosl > 100 ns) + other breakage
[09:00:58] <pcw_home> just verified here but my system is a bit weird so maybe not the best reference
[09:04:17] <cradek> I was unable to get stepconf to do anything that remotely made sense, when configuring my 5i25
[09:04:47] <cradek> oops I mean pncconf
[09:04:52] <skunkworks__> yes - I cannot set the step time/space/dir any higher than 100
[09:04:54] <cradek> stepconf doesn't handle 5i25 of course
[09:05:06] <skunkworks__> well that was your problem!
[09:06:56] <pcw_home> so stepconf is truly busted
[09:07:59] <cradek> argh
[09:08:49] <cradek> so is the image viewer thing on the forum, wow
[09:11:40] <skunkworks__> really? works here in firefox
[09:12:16] <cradek> it sticks the image way down and to the left, so if I scroll way down past the rest of the webpage I can see the right half of the image, and there's no way to see the left half
[09:12:57] <cradek> oh I can "open link in new tab" and avoid the breakage
[09:15:34] <skunkworks__> odd
[09:15:56] <cradek> so looks like all those entries have wrong constraints (<= 100)
[09:16:43] <pcw_home> dir setup time works (oddly enough)
[09:16:45] <cradek> I'm not getting the mm/tour and Pas/mm. is he saying that's unexpected too?
[09:16:47] <skunkworks__> I wonder when that would have been changed..
[09:17:17] <pcw_home> yeah the per axis pages are really messed up also
[09:24:40] <cradek> when it's doing the wrong thing, it spews "update update update" to stdout, so at least it should be easy to find
[09:27:50] <cradek> pcw_home: elaborate on really messed up?
[09:29:11] <cradek> I wonder if this is really stepconf bugs, or more glade/gtk breakage
[09:36:14] <pcw_home> well home latch direction allows selection of sherline,xylotex etc
[09:38:23] <pcw_home> ahh but only only broken on X axis
[09:49:21] <pcw_home> hmm theres only one print update and its in stepconf.py in update_axis_test
[09:54:58] <pcw_home> why the heck is that being called
[09:55:46] <seb_kuzminsky> cradek: i can't fetch from git.linuxcnc.org
[09:57:29] <seb_kuzminsky> might be a local problem
[09:58:37] <seb_kuzminsky> yep, it works fine, my bad, never mind
[10:08:35] <cradek> can anyone tell: is 2.6's stepconf also broken on lucid?
[10:13:28] <cradek> pcw_home: axisx and axisy have separate glade files, but they don't differ in any way that would explain the latch direction being weird on only X.
[10:17:29] <seb_kuzminsky> cradek: i'll test stepconf frmo 2.6.0 on lucid
[10:17:38] <cradek> thanks
[10:21:09] <skunkworks_> https://groups.yahoo.com/neo/groups/mach1mach2cnc/conversations/messages/145043
[10:23:41] <seb_kuzminsky> "AND I will allow you to have the industrial features"
[10:24:31] <cradek> "please pay me to test what I hope to sell for big bucks"
[10:25:38] <cradek> commercial software, how the hell does it work?
[10:25:42] <seb_kuzminsky> heh
[10:26:26] <cradek> this is looking like weird glade/gtk breakage to me :-(
[10:27:18] <seb_kuzminsky> yeah it sure doesn't look like stepconf has change anytime recently
[10:27:41] <cradek> I can't explain the thing with the combobox being broken on the X screen but not the Y screen
[10:27:58] <Tom_itx> cradek, the big boys just have you pay for an annual support fee for fixes
[10:28:11] <cradek> the glade files are the same, and those entries aren't dynamic or anything
[10:28:44] <pcw_home> or test_axis being called when you set the steptime
[10:29:03] <cradek> it's setting the timings for the (appearing later) test axis button to work
[10:29:13] <cradek> I think that's on purpose
[10:29:30] <cradek> not that I traced the code flow
[10:30:29] <pcw_home> It looks like it shoud only be called when test axis is clicked
[10:30:30] <pcw_home> def on_xaxistest_clicked(self, *args): self.a.test_axis('x')
[10:44:09] <seb_kuzminsky> with 2.6.0 on wheezy and precise i cannot change step time from 100, still working on lucid
[10:45:01] <skunkworks_> Mach4 Parallel Port Legacy Addon (cost applicable)
[10:45:40] <seb_kuzminsky> back in a bit
[10:46:29] <seb_kuzminsky> skunkworks_: just saw your message about mach4 on emc-users, was that for steve blackmore? ;-)
[10:46:38] <skunkworks_> and aram
[10:47:30] <skunkworks_> steve stallings has skin in the game.. Maybe some will use his interface board..
[10:48:55] <skunkworks_> I wonder how much the printer port module is...
[10:49:07] <skunkworks_> it doesn't say anywhere
[10:49:37] <Tom_itx> hardware _is_ quite expensive for _industrial_ machines ya know...
[10:50:26] <skunkworks_> http://www.machsupport.com/shop/mach4-hobby/
[10:51:24] <skunkworks_> bottom paragraph
[10:51:25] <archivist> hehe his software is currently non-operational
[10:51:48] <Tom_itx> that's crazy
[10:53:32] <skunkworks_> I wonder what lockup gene is seeing with limit switches
[10:53:51] <Tom_itx> Mach4 is completely new software and has less than 1% of the programming code in common with Mach3.
[10:53:58] <Tom_itx> that's also scary
[10:54:19] <ssi> "It was written from the ground up to be expandable, flexible, and extremely responsive for use with very large files on any operating system; including Windows, Apple, and Linux."
[10:54:27] <ssi> I wonder what it's written in to be that cross-platform
[10:54:28] <cradek> why would you pay to lock the software that doesn't exist yet to a particular pc?
[10:54:44] <cradek> that's extremely silly
[10:54:46] <Tom_itx> and there are no drivers for it yet
[10:54:56] <Tom_itx> but go ahead and pre pay for it...
[10:55:02] <archivist> vapourware
[10:55:04] <ssi> also: the demo download is an exe
[10:55:07] <skunkworks_> people will!
[10:55:08] <ssi> so i call bs :)
[10:55:25] <skunkworks_> ssi, the linux version is in the works..
[10:55:52] <ssi> he's gonna write and maintain three different versions?
[10:56:58] <Connor> wouldn't there be 6 ? Mach4 Lite Mach4 Hobby Mach4 Industrial 1 each for windows 1 each for linux ?
[10:57:23] <ssi> nah the "versions" on a platform are just feature disabling
[10:57:46] <ssi> and it says "Windows, Apple, and Linux"
[10:57:52] <ssi> "Apple" isn't even a platform
[10:59:15] <CaptHindsight> can you control it with your smartphone or tablet while at the beach?
[11:01:15] <skunkworks_> http://www.machsupport.com/shop/mach4-hobby/
[11:01:20] <skunkworks_> oops
[11:01:25] <skunkworks_> Mach4 is 100% done in C++, Graphics done with wxWidgets, We used Dialog Blocks for the Dialog designer..
[11:01:58] <CaptHindsight> Mach4 will pick up users that just want to install and have a bunch of graphical config options to choose from, no looking at or editing config files
[11:03:41] <skunkworks_> a) why would hitting a limit switch lock things up and b) why would you hit a limit switch? (when you have soft limits)
[11:03:53] <CaptHindsight> those users want the software to detect and configure everything
[11:05:10] <cradek> skunkworks_: in 2.6.0 on my little lathe, the limit switch works fine. I poked it to make sure, before homing the first time (since it's home+limit right before a hard stop)
[11:05:52] <skunkworks_> I wonder what genes problem is.. He should post his hal files.
[11:06:18] <skunkworks_> cradek, does the lathe still work?
[11:07:03] <seb_kuzminsky> wheezy & precise has python-glade2 version 2.24, lucid has 2.17
[11:07:49] <cradek> skunkworks_: yep, I'm currently setting up a wheel/switches so I can use touchy with it
[11:08:34] <cradek> seb_kuzminsky: I wonder if opening and saving the files in glade, or some other crazyass voodoo, would make it work right again
[11:08:37] <skunkworks_> cradek, nice!
[11:10:32] <seb_kuzminsky> when i run the 2.6.0 stepconf on lucid it pre-populates the Step Tiem field with 5000 (it's 100 on precise & wheezy)
[11:10:56] <cradek> does the latch direction thing on the X axis page contain the right things?
[11:11:14] <seb_kuzminsky> and i can change it up to 100000 fine, seems to work fine
[11:11:20] <cradek> I think it's supposed to be same and opposite
[11:12:01] <seb_kuzminsky> it's grayed out, but it says Same
[11:12:03] <cradek> you have to put home switches on parport pins to activate it
[11:12:08] <seb_kuzminsky> ah
[11:12:33] <seb_kuzminsky> yah, Same and Opposite
[11:12:41] <cradek> argh, I was afraid of this
[11:12:42] <seb_kuzminsky> seems totally fine, afaics
[11:12:45] <seb_kuzminsky> yeah
[11:12:53] <cradek> fffff
[11:13:08] <seb_kuzminsky> i gotta pay attention to my dayjob, i'll be back tonight
[11:14:39] <cradek> thanks for testing
[11:15:27] <seb_kuzminsky> the folks at #pygtk (on irc.gnome.org) might have advise
[11:15:32] <seb_kuzminsky> http://www.pygtk.org/feedback.html
[11:15:33] <seb_kuzminsky> bbl
[11:15:57] <jepler> "simplify your application until it's a reasonable test case, then ask again"
[11:16:10] <jepler> (not snark about past treatment on #pygtk, just what I'd ask for myself)
[11:16:44] <skunkworks_> that sounds farmiliar...
[11:16:47] <skunkworks_> ;)
[11:18:29] <jepler> will someone make there be an issue on sf for this? I might look at it this weekend too, and it would be great to have a more authoritative statement of the known problem(s) without having to read back through irc
[11:18:41] <jepler> .. I came in on the tail end, it looks like
[11:19:42] <cradek> I will do it
[11:19:58] <jepler> thank you!
[11:23:06] <cradek> I need to play with a kitten
[11:27:28] <jepler> that sounds like fun
[11:45:59] <cradek> https://sourceforge.net/p/emc/bugs/385/
[11:46:13] <cradek> I think this is a correct summary. pcw_home, seb_kuzminsky: please add to it if I missed something
[11:58:04] <jepler> cradek: thank you. i understand the HTR and reproduce those behaviors.
[12:04:41] <cradek> I'm sad that people on the forum keep giving the url of the new iso, instead of the announcement that says all the things about how to use it
[12:05:39] <cradek> alex_joni: what are your feelings about changing the "Download" page - removing all the old stuff and putting the new instructions there?
[12:22:30] <Tom_itx> removing 8.xx? some (me) still use that on one machine
[12:22:38] <Tom_itx> and if i ever lost the disk...
[12:24:21] <jepler> hm, glade is not fast at loading things
[12:24:59] <jepler> but I think I have a partial answer to what is going on with stepconf
[12:25:07] <cradek> ooh
[12:25:09] <jepler> it has been split into multiple glade files
[12:25:30] <jepler> glade file "main_page.glade" and "base.glade" both have an "adjustment1"
[12:26:04] <jepler> on "main_page.glade", adjustment1 has a range from 1 to 100
[12:26:11] <jepler> somehow that is used on page "base.glade"
[12:27:17] <jepler> something similar is happening for "liststore1", used in "base.glade" for driver type and in "axisx" / "axisy" / "axisz" for latch type
[12:27:43] <jepler> if I rename "adjustment1" in "base.glade" to "baseadjustment1", it fixes the step timing field
[12:28:16] <jepler> if I rename "liststore1" to "axisxliststore1" in "axisx.glade", it fixes the combobox for X axis, but now the Y axis (which is still "liststore1") is wrong
[12:28:26] <cradek> cool
[12:28:32] <jepler> so basically .. when things have the same name in multiple glade files, it breaks in weird ways
[12:28:44] <cradek> so there is some behavior change or new bug about loading glade files
[12:28:47] <jepler> and I think it can be cured by making sure they have distinct names
[12:29:21] <jepler> bbl, lunchtime
[12:39:20] <seb_kuzminsky> awesome
[12:39:26] <seb_kuzminsky> 2.6.1, here we come :-)
[12:46:10] <Tom_itx> heh
[12:51:01] <cradek> "You should be sure that there is no conflict with names in the two files"
[12:51:10] <cradek> says some guy, http://stackoverflow.com/questions/1378582/using-multiple-glade-files-gtkbuiler#1380024
[12:55:52] <seb_kuzminsky> there are a ton of duplicates:
[12:56:08] <seb_kuzminsky> src/emc/usr_intf/stepconf> grep 'object.*id="' *.glade | sed -re 's/^.*id="([^"]*)".*$/\1/' | sort | uniq --repeated | wc -l
[12:56:11] <seb_kuzminsky> 86
[12:57:24] <skunkworks_> oops
[13:02:12] <jepler> well seb's script is sure simpler than mine!
[13:02:34] <seb_kuzminsky> except for that burst of line noise in the middle ;-)
[13:03:15] <cradek> sed probably isn't the best tool to use to parse xml...
[13:03:40] <seb_kuzminsky> heh
[13:04:41] <seb_kuzminsky> 2.6.0 may be ths shortest-lived linuxcnc release ever
[13:04:49] <seb_kuzminsky> i should have read #380 before releasing...
[13:05:32] <ssi> .0 releases always have teething issues :)
[13:05:47] <cradek> I think we've had shorter, it's perfectly ok
[13:05:48] <archivist> at least you can point at the bugfix speed in 2.6.1
[13:06:08] <seb_kuzminsky> archivist: good point
[13:07:00] <cradek> jepler: perhaps use a separate builder for each "tab"?
[13:12:57] <seb_kuzminsky> so does it work on lucid because of some change in how the .glade files are loaded?
[13:13:14] <seb_kuzminsky> maybe they're loaded in parallel in later versions?
[13:15:01] <cradek> the call we have is "add this xml tree to the xml tree we have so far" and we do it for each page. somehow it worked with duplicate names before and now it doesn't - it's always built just one tree
[13:15:39] <cradek> each xml file somehow does not need to be in order - definitions before references - in order to work
[13:16:06] <cradek> and it sounds like that situation is what's messing it up
[13:29:48] <seb_kuzminsky> i tried loading a couple of .glade files with colliding object ids into the same gtk.Builder, and it worked without warnings or anything
[13:29:57] <seb_kuzminsky> it's strange to me that it doesn't complain
[13:31:36] <seb_kuzminsky> this says gtkbuilder is supposed to warn about id collissions: https://developer.gnome.org/gtk2/stable/gtk-migrating-GtkBuilder.html
[13:57:01] <cradek> I can't fathom why rick wants the toolno+10000, considering he only wants the one offset
[13:57:30] <cradek> I thought the whole point of this ridiculousness was to get separate geometry and wear offsets
[13:57:41] <cradek> he could have always just used remap
[14:37:30] <memleak> with more and more distros using python 3 as the system default, are there any plans on converting linuxcnc to python 3?
[14:38:29] <cradek> I bet 2 and 3 will coexist for a really long time
[14:39:22] <memleak> the reason i ask is so users dont need to switch to python 2 for system default or use --with-python configure vars
[14:39:57] <cradek> I'm more worried about ongoing gnome/gtk breakage
[14:40:19] <cradek> that's much more breakage-prone than python, and python is much more breakage-prone than tcl
[14:41:10] <cradek> also we'd break compatibility with older python-2-only systems, and we must only do that for a good reason
[14:41:32] <memleak> ah right. compatibility. forgot about that.. i do that a lot xD
[14:41:46] <cradek> :-)
[14:47:56] <memleak> does linuxcnc work with gold (binutils) or just bfd?
[14:50:19] <cradek> jepler, seb_kuzminsky: http://pygtk-id.blogspot.com/2009/06/replace-duplicate-widget-name-in-glade.html
[14:55:35] <jepler> memleak: I frequently build linuxcnc on a system with $ ld --version
[14:55:35] <jepler> GNU gold (GNU Binutils for Debian 2.22) 1.11
[14:55:43] <memleak> jepler, nice thanks!
[14:56:10] <memleak> jeeze my sound card just caused a kernel panic..
[14:56:27] <memleak> creative drivers really are as bad as people say.
[14:56:41] <memleak> and the windows ones arent any better..
[14:59:41] <jepler> to the best of my recollection, I've never bought an add-on soundcard since they became commonplace on motherboards -- early in the ATX era?
[15:00:14] <cradek> I didn't know that was still a thing people did
[15:00:51] <memleak> my external sound card has a quad core CPU embedded on it..
[15:01:46] <skunkworks_> for that real hard hitting base?
[15:01:46] <memleak> if your a gamer audio is pretty important.
[15:01:48] <skunkworks_> bass
[15:02:03] <memleak> audio position / HRTF is what its for.
[15:04:08] <jepler> "Head-related transfer function" sounds like it is going to end up being a joke like PEBCAK
[15:04:11] <jepler> but I guess not
[15:09:20] <seb_kuzminsky> cradek: that blogspot script looks like a good start, but it would make really confusing ids for stepconf, since stepconf's ids already end with numbers
[15:24:40] <seb_kuzminsky> cmorley: have you seen the talk about glade problems on wheezy? you probably know glade best of anyone here, do you have any suggestions?
[15:32:01] <memleak> heh jepler..
[15:46:30] <jepler> so I now have my hands on an Odroid U3 running their distro based on Ubuntu 14.04. It builds master branch, but unfortunately it takes longer than the arm chromebook (7 minutes vs 5:30).
[15:46:57] <jepler> linuxcnc does run, but as it's not a PREEMPT-RT kernel the result is not worth reporting
[15:47:24] <jepler> unfortunately, the ethernet is connected internally via USB, so it's not useful for playing with 7i80 in realtime even if I had a PREEMPT-RT kernel.
[15:48:23] <seb_kuzminsky> jepler: are you building with -j6?
[15:48:30] <jepler> seb_kuzminsky: I found that -j4 was faster than -j5
[15:48:33] <jepler> -j5 was 10 minutes
[15:49:43] <jepler> it's got the fan+heatsink combo mounted, and the fan doesn't even stay on at all tim esduring the build, so it shouldn't be thermal throttling
[15:49:56] <jepler> s/tim esduring/times during/
[15:50:20] <seb_kuzminsky> ah ok
[15:51:01] <jepler> Runtest: 142 tests run, 142 successful, 0 failed + 0 expected
[15:57:22] <seb_kuzminsky> jepler: my u3 just built master in 5:01 wall with -j6
[15:57:37] <PCW> wonder how this would do:
[15:57:38] <PCW> https://developer.nvidia.com/jetson-tk1
[15:57:50] <seb_kuzminsky> that's running wheezy on some kind of fancy fast micro-sd card
[15:58:53] <seb_kuzminsky> err, my u3 is running wheezy, i mean
[15:59:12] <seb_kuzminsky> PCW: do you know what they mean by "4-plus-1 quad-core ARM Cortex A15 CPU"?
[15:59:34] <PCW> 4 A15s and a A7
[15:59:38] <seb_kuzminsky> ah
[15:59:52] <seb_kuzminsky> the u3 has 4xA9, clocked at 1.7 Ghz
[16:00:26] <seb_kuzminsky> they both have 2 GB RAM
[16:02:01] <micges_garage> PCW: finally moved, I'll get to hm2_eth 4 packet problem in few hours
[16:02:13] <seb_kuzminsky> i like that the jetson has a normal atx power connector and a pcie slot
[16:03:00] <PCW> well miniPCIE
[16:04:13] <PCW> the molex power conn is for a SATA drive (the card runs on 12V)
[16:04:40] <seb_kuzminsky> oh
[16:04:44] <seb_kuzminsky> then i like it less ;-)
[16:06:44] <seb_kuzminsky> oh yeah, -j4 took 4:48, quicker than -j6
[16:10:12] <jepler> seb_kuzminsky: huh, I wonder why mine's so much slower!
[16:11:01] <seb_kuzminsky> i have no fan on mine, maybe your fan blows away some of the cycles
[16:11:26] <seb_kuzminsky> what do you have for disk?
[16:11:49] <jepler> seb_kuzminsky: their eMMC, 16GB
[16:11:54] <jepler> seb_kuzminsky: what OS are yo uon?
[16:13:02] <seb_kuzminsky> wheezy (7.6)
[16:13:15] <seb_kuzminsky> make -j6 with a hot cache took 4:49
[16:15:08] <jepler> seb_kuzminsky: and kernel?
[16:15:29] <jepler> with ccache installed, the second build from clean is just 1:05
[16:16:08] <seb_kuzminsky> Linux wheezy-armhf-u3 #1 SMP PREEMPT Sat Jun 22 00:52:21 BRT 2013 armv7l GNU/Linux
[16:16:21] <jepler> Linux odroid #1 SMP PREEMPT Tue Jun 10 14:54:36 UTC 2014 armv7l armv7l armv7l GNU/Linux
[16:17:15] <seb_kuzminsky> mmm, ccache
[16:19:20] <jepler> anyway, it's enough faster than a bbb that it wouldn't hurt to develop linuxcnc on it
[16:22:28] <seb_kuzminsky> yeah, the bbb is painfully slow
[16:22:33] <seb_kuzminsky> the u3 is totally usable
[16:22:49] <jepler> seb_kuzminsky: is your /usr/bin/gcc the binary, or is it the perl script /usr/bin/hardened-cc?
[16:23:02] <jepler> gcc version and the presence of hardening wrappers could be two other reasons for differning performance
[16:23:25] <seb_kuzminsky> it's a symlink to gcc-4.6, which is a binary
[16:23:33] <jepler> gcc 4.8 here
[16:23:53] <jepler> I could install gcc-4.6 and configure with it, I suppose
[16:24:00] <jepler> .. I'll probably DL a wheezy image this weekend though
[16:24:40] <jepler> seb_kuzminsky: is your board the newer rev with SPI header J5?
[16:24:50] <seb_kuzminsky> i bought the u3 to be a linuxcnc buildslave, it's weird having a buildslave that's actual hardware instead of oa vm
[16:25:03] <seb_kuzminsky> i keep waiting for a nice arm virtualization host
[16:25:12] <seb_kuzminsky> mine's the old (0.2?) one without spi
[16:25:25] <jepler> qemu-arm not performant enough?
[16:25:30] <seb_kuzminsky> not near
[16:25:54] <seb_kuzminsky> i tried that first but gave it up because it's too slow, slower than a bbb even
[16:26:01] <seb_kuzminsky> on my fairly fast amd64 machines
[16:26:10] <jepler> yuck
[16:26:36] <jepler> have you ever worked out how to use ccache in your buildbot builds? we use it at $DAY_JOB, but that's without pbuilder as a complicating factor.
[16:27:53] <seb_kuzminsky> pbuilder integrates with ccache, i use it that way on the rtai/linux builds i did
[16:28:00] <seb_kuzminsky> http://www.amd.com/en-us/press-releases/Pages/64-bit-developer-kit-2014jul30.aspx
[16:28:52] <jepler> oh only $3k
[16:29:02] <jepler> :-/ you can buy one of each fly-by-night arm board for that kind of dough
[16:29:52] <seb_kuzminsky> yeah, but the a57 does hardware virtualization
[16:29:56] <seb_kuzminsky> (still not worth it)
[16:30:21] <seb_kuzminsky> i should probably enable ccache in the buildbot some day
[16:30:33] <seb_kuzminsky> i wonder if it works on docs, that's what takes the most time
[16:30:39] <jepler> no :(
[16:30:44] <jepler> that would be too good to be true
[16:31:07] <jepler> in your place I'd jump up and down for a debian/configure option to turn off doc building, and only do it on a minimum of machines
[16:31:25] <jepler> (I sort of thought we'd added one, but it wasn't there when I looked)
[16:32:26] <seb_kuzminsky> i'd rather wait longer and know that the docs build on all our supported platforms
[16:33:01] <CaptHindsight> seb_kuzminsky: what is it you like about the A57's from AMD?
[16:33:10] <jepler> CaptHindsight: 16:11:16 <+seb_kuzminsky> yeah, but the a57 does hardware virtualization
[16:34:47] <CaptHindsight> jepler: I noticed that comment, I was wondering besides that. I should have been more clear.
[16:35:12] <jepler> ah well I think that's the thing
[16:35:57] <seb_kuzminsky> that's it - i just want to do our armhf builds in vms instead of on aactual hardware
[16:37:04] <jepler> grrrrr
[16:37:12] <jepler> this device has stubbornly been saying
[16:37:15] <jepler> Sat Aug 2 06:48:00 CST 2014
[16:37:23] <jepler> but the time in CST is of course 16:18
[16:37:27] <CaptHindsight> few seem to like or use crosstool-ng
[16:37:34] <jepler> and I'm doing things like running ntpdate
[16:37:38] <jepler> ntpdate[31461]: adjust time server offset -0.004951 sec
[16:37:41] <jepler> nope, still wrong
[16:37:46] <jepler> $ cat /etc/timezone
[16:37:46] <jepler> Australia/Adelaide
[16:37:57] <jepler> seems there's more than one CST in the world :-/
[16:38:26] <jepler> seb_kuzminsky: with hardening-wrapper removed, gcc-4.6, -j4, no ccache, hot disk: 4 minutes
[16:38:36] <seb_kuzminsky> nice!
[16:38:42] <jepler> so now it is my fastest ARM machine
[16:39:08] <seb_kuzminsky> is your the new 0.5 version with the spi header?
[16:39:24] <seb_kuzminsky> *yours
[16:39:48] <jepler> seb_kuzminsky: yes
[16:40:05] <seb_kuzminsky> 4xA9 @ 1.7 GHz still?
[16:40:13] <jepler> yes
[16:40:42] <seb_kuzminsky> i guess the eMMc is responsible for the speedup (i run on micro-SD)
[16:41:32] <jepler> so now I need just 4 more things: preempt-rt kernel with good performance; a mesa card with SPI firmware; a 1.8v level translator for the SPI communications; and a hm2_spi driver
[16:42:09] <jepler> anyway, bbl.
[16:42:14] <jepler> seb_kuzminsky: thanks for comparing numbers with me
[16:42:14] <seb_kuzminsky> seeya
[16:42:21] <seb_kuzminsky> always fun
[16:44:40] <jepler> seb_kuzminsky: .. or you could be seeing a modest amount of thermal throttling
[16:47:43] <seb_kuzminsky> i think i threw a towel over my u3 to protect it from chips last time i ran my milling machine...
[17:49:07] <seb_kuzminsky> gscreen's glade files duplicate label1 and table1
[17:51:20] <seb_kuzminsky> gscreen, gmoccapy, and gladevcp all deliberately load user-specified glade files, it'd be nice if we could detect & warn about id collisions, instead of just silently doing the wrong thing like now
[18:38:50] <seb_kuzminsky> gtk.Builder() only detects duplicated object ids if they're in the same .glade file
[18:53:30] <andypugh> linuxcnc.org “download” header takes you to 2.5. Is there a 2.6 LiveCD?
[18:58:32] <seb_kuzminsky> there is, but i think cradek is still considering it experimental
[18:58:52] <seb_kuzminsky> http://article.gmane.org/gmane.linux.distributions.emc.user/52401
[19:00:35] <seb_kuzminsky> hm, alex_joni said he fixed the wlo webpage links
[19:01:15] <seb_kuzminsky> ah, he added the 2.6.0 release to the News section but didn't update the Download page
[19:03:30] <seb_kuzminsky> andypugh: i don't have access to the wlo webpage, so i opened a bug, hoping that someone else will fix it
[19:04:32] <andypugh> It is a bit of a problem we ought to address, that nobopdy knows who can edit the web page.
[19:05:15] <andypugh> I think I might be able to, but I am scared of the consequences of messing up.
[19:08:18] <andypugh> it seems that I _can_ edit this page: http://www.linuxcnc.org/administrator/index.php?option=com_content&sectionid=-1&task=edit&cid[]=2
[19:08:32] <andypugh> But I don’t know wjat to edit it to…
[19:29:00] <andypugh> I put an initial placeholder at http://www.linuxcnc.org/index.php/english/download
[19:29:23] <andypugh> The Cabal need to approve new content
[19:34:23] <andypugh> While we are at it, this seem to be an orphan link, I don’t see how to get there from the front page. http://www.linuxcnc.org/index.php/english/component/content/article/17-names-2002