Back
[05:15:52] <REEEN> hello, is someone of you familiar with virtual box, I'm trying to use linuxcnc under windows for developing some stuff
[09:06:28] <skunkworks> linuxcnc is down.
[09:06:31] <skunkworks> ;)
[09:18:29] <seb_kuzminsky> :-/
[09:20:15] <skunkworks> seb_kuzminsky, how goes the summer? stella found out she likes going down water slides..
[09:42:56] <mozmck> So who can update Joomla on the website?
[09:43:01] <mozmck> SWPLinux: hi!
[09:43:04] <SWPLinux> that would be me
[09:43:07] <SWPLinux> hi mozmck
[09:43:14] <mozmck> long time no see
[09:43:17] <SWPLinux> yep
[09:43:18] <skunkworks> oh my - 2 days in a row!!
[09:43:39] <SWPLinux> well, the website is broken :)
[09:43:47] <skunkworks> you are a saint!
[09:44:17] <skunkworks> are you still in WI?
[09:44:18] <SWPLinux> There's a "finish upgrade" button on the DreamHost panel. You now know everything I know about upgrading Joomla :)
[09:44:25] <SWPLinux> yep, in Manitowoc now
[09:44:31] <skunkworks> nice
[09:44:50] <mozmck> haha! what version of Joomla does the site use now?
[09:45:23] <SWPLinux> I haven't done any database dump, that's the only thing I'm concerned about at this point - it's not like I can break it any worse than it is already
[09:45:35] <SWPLinux> cradek: do your backups include SQL dumps?
[09:45:54] <SWPLinux> mozmck: I'll check, give me a second to log in and stuff
[09:45:58] <mozmck> What it might do is break plugins if the site uses any.
[09:46:32] <SWPLinux> well, since the whole site is broken, I stand by my assertion that it's hard to break it any worse :)
[09:46:38] <mozmck> We tried to update a site from 2.5 to 3.x, but some plugins did not support 3.x yet (VirtueMart was one I think).
[09:47:02] <SWPLinux> luckily we don't sell anything
[09:47:12] <mozmck> yep
[09:48:17] <mozmck> I think the forum is a plugin, and if that one has a compatible upgrade then you're probably fine.
[09:48:40] <SWPLinux> it's running 1.5.26, the upgrade is to 3.4.1 (or at least that's the version they show in the one-click install list)
[09:49:05] <mozmck> Oh, wow, that's a little old :)
[09:49:16] <SWPLinux> yep, that's why they disabled it
[09:49:23] <SWPLinux> security concerns and all
[09:49:47] <SWPLinux> I wonder how hopeful I should be that the database will be migrated automatically
[09:50:37] <mozmck> Upgrades I have done have usually updated the database, but I haven't used Joomla much.
[09:51:04] <mozmck> You should be able to do a backup of the database from the Dreamhost control panel.
[09:51:55] <cradek> SWPLinux: yes: -rw------- 1 chris chris 90526183 Jul 8 23:20 linuxcnc_org.sql
[09:53:27] <SWPLinux> ok, cool
[09:53:57] <SWPLinux> There are automatic backups from DH, so I can restore to several versions from at least the last week
[09:55:59] <cradek> SWPLinux: thanks for handling this
[09:56:15] <SWPLinux> thank me after it's handled :)
[09:58:49] <seb_kuzminsky> SWPLinux: thanks for working on it :-)
[10:33:56] <skunkworks> SWPLinux, still doing camera stuff?
[10:34:18] <SWPLinux> yep
[10:34:40] <skunkworks> neat - how is it going?
[10:39:52] <SWPLinux> they're doing about 100 jobs a year. I don't travel with it much these days
[11:24:23] <mozmck> Where is the code for the linuxcnc python module? I found it once and lost it again.
[11:34:34] <mozmck> I think I found it in src/emc/usr_intf/axis/extensions/emcmodule.cc
[11:38:25] <seb_kuzminsky> mozmck: yeah that's it
[11:38:29] <seb_kuzminsky> funny place for it
[11:38:57] <mozmck> Axis was probably the only thing that used it at first.
[11:39:25] <mozmck> maybe it should be moved though?
[11:48:55] <seb_kuzminsky> yeah, i think that'd be a good idea (in master)
[12:22:45] <seb_kuzminsky> linuxcnc.org is back
[12:23:12] <cradek> I haven't heard yet that it's done
[12:23:13] <jepler> so I have a distinct memory of a bug affecting ARM systems that was related to casting between floating point types and unsigned types
[12:23:26] <jepler> .. and I think this bug is fixed at least in 2.7, if not in 2.6 and earlier
[12:23:33] <jepler> but I can't find any commit about it
[12:23:33] <cradek> stupidly, it has to be enabled to upgrade it, so he may still be working
[12:23:54] <jepler> the symptom of this may have been that "motors would only turn one way"
[12:25:39] <jepler> anybody else have a bell rung by that?
[12:31:11] <seb_kuzminsky> not i
[12:31:47] <cradek> I don't see anything either
[12:33:53] <seb_kuzminsky> i sometimes recall writing some code only to realize it's on some forgotten unmerged local branch
[12:34:13] <cradek> yeah I scanned origin/jepler/... branches with no luck
[12:34:24] <cradek> are you having the same problem again or something?
[12:37:53] <jepler> I discussed it on irc 9/3 and 9/4/2014
[12:37:57] <jepler> but I never see anything pushed
[12:38:06] <jepler> somebody e-mailed me asking about it
[12:38:15] <jepler> for some reason I looked into it instead of ignoring their message
[12:38:39] <jepler> Reflog message: commit: hm2: avoid losing negative velocity commands on arm
[12:39:28] <jepler> hm look, I have a whole branch called "jepler/proposed-master" on this system
[12:41:11] <jepler> fe6fba2 configs: sample configuration for hm2_spi
[12:41:11] <jepler> af38a89 hm2: avoid losing negative velocity commands on arm
[12:41:15] <jepler> these two are the unpushed ones
[12:43:17] <seb_kuzminsky> those sound useful :-)
[12:44:54] <skunkworks> that sounds like the issue someone was having on the rpi
[12:44:59] <skunkworks> (MK)
[12:45:09] <jepler> seb_kuzminsky: I'm not in a position to check if my hm2_spi / stepgen configuration actually works. should I push it to 2.7 anyway?
[12:45:35] <jepler> I'm more comfortable not pushing it
[12:45:41] <jepler> let me just push the bit I'm confident about
[12:46:15] <KGB-linuxcnc> 03Jeff Epler 052.7 8b536f4 06linuxcnc 10src/hal/drivers/mesa-hostmot2/stepgen.c hm2: avoid losing negative velocity commands on arm * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=8b536f4
[12:49:00] <seb_kuzminsky> if you thin kthe spi config is a better starting point than no config at all, then push it
[12:49:55] <seb_kuzminsky> wow, that's weird-looking patch
[12:50:09] <jepler> I don't think there's anything magic about the spi config
[12:50:23] <seb_kuzminsky> i think if i saw that my instinct would be to remove the (int32_t)
[12:50:30] <jepler> I think it's a matter of setting a few ini items
[12:50:30] <jepler> +[HOSTMOT2]
[12:50:30] <jepler> +DRIVER=hm2_spi
[12:50:30] <jepler> +BOARD=7i90
[12:50:42] <seb_kuzminsky> because all the world's an x86
[12:51:34] <jepler> I'll add a comment
[12:52:29] <seb_kuzminsky> thx
[12:52:40] <seb_kuzminsky> future seb thanks you
[12:53:06] <jepler> // the double cast here is intentional. (uint32_t)(-1.0) is undefined in
[12:53:10] <jepler> // C (and in practice it gives the undesired value 0 on arm systems), but
[12:53:13] <jepler> // (uint32_t)(int32-t)(-1.0) is defined and gives the desired value on all
[12:53:16] <jepler> // systems.
[12:53:19] <seb_kuzminsky> awesome
[12:53:47] <seb_kuzminsky> i'm rarely disciplined enough to figure out why wonky-looking code is wonky-looking by reading the logs
[12:53:54] <KGB-linuxcnc> 03Jeff Epler 052.7 2b5258c 06linuxcnc 10src/hal/drivers/mesa-hostmot2/stepgen.c hostmot2: comment on odd-looking code * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=2b5258c
[12:54:57] <jepler> ho boy it looks like a PITA to test kernels on the dragonboard
[12:55:19] <seb_kuzminsky> there must be some kind of static analyzer that can tell us everywhere we do things that are undefined in C
[12:55:47] <jepler> (uint32_t)(1.0) is perfectly well defined so you have to know where the runtime range will include negative numbers
[12:56:00] <seb_kuzminsky> in this case you could know it at compile time
[12:58:25] <seb_kuzminsky> and i guess if the analyser couldnt be confident, then the human couldnt either, and there should be explicit range checks added before the cast
[12:58:39] <jepler> if (new_vel > maxvel) {
[12:58:39] <jepler> new_vel = maxvel;
[12:58:39] <jepler> } else if (new_vel < -maxvel) {
[12:58:39] <jepler> new_vel = -maxvel;
[12:58:39] <jepler> }
[12:59:02] <jepler> I can see that a static analyzer fed this fragment should assume new_vel could be negative
[13:00:08] <cradek> jepler: f8e7f6728 ?
[13:01:34] <jepler> cradek: huh I don't remember that at all, I wonder what behavior it led to
[13:02:04] <cradek> did you see the previous one too?
[13:02:50] <jepler> cradek: I did find the patch I was looking for -- it wasn't on g.l.o yet
[13:02:56] <cradek> oh ok
[13:04:26] <jepler> thanks for going diving!
[13:04:35] <cradek> it was fun
[13:35:03] <jepler> hm I should have lunch
[13:35:37] <jepler> my self-built kernel boots, spews some messages too fast for me to see, then clears the screen and apparently locks up
[13:38:28] <skunkworks> sucess! Fail!
[13:39:07] <cradek> can it do serial console?
[13:42:04] <jepler> cradek: I think there's a serial console but if so it's on the 2mm header and last time I went poking jumpers into one I ruined the connector
[13:42:34] <jepler> I would probably do OK if I did wires rather than those blocky jumpers intended for .1" pins
[13:43:05] <cradek> oh yes that ended with much gnashing of teeth and soldering
[13:47:20] <jepler> yay I got the driver built for this usb wifi dongle at least
[13:47:29] <jepler> though for some reason the wifi is behaving better today
[14:06:23] <kwallace_shiz> Hello. I'm finally starting to try remapping on my HNC lathe. Currently I'm using an M10x.ngc with "halcmd setp ..." to set/unset the appropriate pin signals for things like the spindle high/low clutch. I'm wondering what the most direct method would be with remapping?
[14:06:58] <cradek> that's how I did it too. what about the behavior do you want to improve?
[14:10:35] <kwallace_shiz> The plan is to use M41 for "engage low" which is easier (?) to remember. Actually, M3 P1 for forward, first gear might be easier. Too many options.
[14:15:21] <kwallace_shiz> With remapping it seems there are three program options: 1) call an ngc file, 2) call a python file and which calls g-codes, 3) call a python file which calls canon somethings. Using canon seems to be the most direct, but I don't know if canon and set pins.
[14:16:05] <kwallace_shiz> s/ and / can
[14:16:07] <cradek> I guess you could make M41 call M101, but why bother?
[14:17:46] <kwallace_shiz> I often need to scan my M10x files to figure out which one does what I want.
[14:22:33] <SWPLinux> so, I think I'm at the point where I can upgrade joomla. does anyone think it's a good idea for me to do that?
[14:22:39] <SWPLinux> or a bad idea?
[14:23:15] <jepler> what's the alternative?
[14:23:26] <SWPLinux> not upgrading
[14:23:45] <SWPLinux> I don't know if DH will turn it off again
[14:23:46] <jepler> just turning back on the version that dreamhost disabled because it is insecure, and leaving it that way? that doesn't seem a great solution.
[14:24:11] <jepler> I think upgrade it
[14:24:20] <SWPLinux> I think so too
[14:24:23] <cradek> I sure think that's the thing to try
[14:24:33] <SWPLinux> I just hope nothing serious goes wrong
[14:24:36] <jepler> if something goes wrong we should be able to put back one of cradek's backups and try again
[14:24:55] <SWPLinux> at least it looks like jUpgrade makes a copy in a subdirectory, it might not break anything even if it doesn't work
[14:25:31] <jepler> .. do you ever just hate all software?
[14:25:39] <SWPLinux> well, here goes. I wonder if it'll be noticeable
[14:26:01] <SWPLinux> yes, usually when using something like MS Project or any graphics editor
[14:26:03] <cradek> sometimes I briefly don't hate a certain piece of software
[14:26:21] <jepler> I hardly ever hate "true", I guess.
[14:26:40] <jepler> "cat", "mv", and even "ls" have made my s--t list from time to time
[14:26:55] <SWPLinux> ls does have a lot of options
[14:27:14] <SWPLinux> false can be good, if you're in the right mood
[14:28:15] <SWPLinux> well, I clicked the big button. now to drool while it works
[14:29:12] <SWPLinux> so, how's the CNC world these days?
[14:30:48] <jepler> oh, we have a new trajectory planner, and run on PREEMPT-RT pretty good. more mesa cards are supported, including ethernet ones.
[14:31:07] <SWPLinux> I've seen a lot of chatter about ARM boards, I think
[14:31:42] <jepler> most people using ARM boards are on the machinekit fork, though.
[14:32:00] <SWPLinux> oh. In other news, there's a machinekit fork :)
[14:32:15] <jepler> yes, that happened about 2 years ago
[14:35:31] <jepler> we weren't able to work out how to get the good parts of their branch into linuxcnc, and there were parts we simply wouldn't accept into linuxcnc
[14:36:03] <jepler> so unfortunately the projects went separate ways, with us announcing the stable 2.6.0 and them announcing machinekit as a separate product at about the same time
[14:39:13] <SWPLinux> is this thing working?
[14:39:16] <SWPLinux> ok, now it is
[14:39:38] <SWPLinux> the upgrade seems to have almost worked. the content is there, but the style/design is wrong
[14:39:53] <SWPLinux> it's at
http://ww.w.linuxcnc.org/jupgrade
[14:42:21] <jepler> and the forum link doesn't work
[14:43:51] <jepler> the first issue is probably just not including the right css
[14:43:53] <jepler> the second issue, no idea
[14:49:09] <jepler> it could be that /jupgrade doesn't have a copy of the forum's php files
[14:50:05] <SWPLinux> possible
[14:50:26] <SWPLinux> if my ssh connection would keep working, I'd look through the directory structure ...
[14:55:25] <SWPLinux> strange. there's a template called lilnuxcnc-float which got migrated over, but it doesn't show up in the template list in admin
[15:01:14] <SWPLinux> ok, getting closer. now it looks more or less right but doesn't have the links
[15:13:29] <SWPLinux> bummer. it looks like kunena didn't get upgraded, even though it's specifically listed as a supported module in the upgrade program
[15:23:36] <SWPLinux> unfortunately, I can't spend too much more time trying to get the upgrade to work
[15:23:54] <SWPLinux> here are the things I know are wrong at this point:
[15:24:32] <SWPLinux> the LinuxCNC style may not have been imported correctly, or there may be some incompatibilities with Joomla 2.5.x
[15:25:20] <SWPLinux> the kunena component didn't get transferred/installed, so the kunena menus and whatever else kunena manages isn't working
[15:25:58] <SWPLinux> the CB (Community Builder) module didn't get transferred/installed, so whatever it manages isn't working (forum, I suspect)
[15:26:38] <SWPLinux> other, less important things that didn't transfer are the geotagging component and the polling component
[15:27:33] <SWPLinux> also, the only language that is installed is English-UK. I think this is for words that Joomla provides (like the copyright stuff and that kind of thing), not the content itself
[15:27:46] <SWPLinux> there may be other things that I haven't noticed yet.
[15:27:57] <jepler> SWPLinux: thanks for doing what you could
[15:28:24] <jepler> I wish we had someone with a clue about php who knew how our website was put together :-/
[15:31:26] <jepler> dangit I still have no clue what is going wrong with my self-built kernel
[15:31:47] <jepler> I guess I need my serial console so I can actually record the messages that occur before it crashes
[15:32:03] <jepler> I had a brainstorm that the crash might be something when userspace reaches the point where it starts X
[15:32:06] <SWPLinux> oh, thanks for reminding me - there's also a php temp dir setting that's wrong (it
[15:32:18] <jepler> .. so after disabling that, my self-built kernel just reboots (into the default kernel)
[15:32:20] <SWPLinux> it's in the warnings in the admin panel
[15:32:50] <SWPLinux> jepler: is this on a PC or some embedded thing (I saw some ARM comments earlier)
[15:38:28] <jepler> SWPLinux: the latter. dragonboard 410c, a sub-$100 "aarch64" (64-bit arm) board
[17:27:51] <PetefromTn_> Well folks making some progress on my anodizing experiments LOL
[17:28:36] <PetefromTn_> Got some inexpensive candy glass thermometers that are good to 300 degrees
[17:28:47] <PetefromTn_> cut some 1/4 inch 6061 plate for my cathodes
[17:29:04] <PetefromTn_> drilled and tapped and attached a ground strap between the two plates
[17:29:43] <PetefromTn_> grabbed a hunk of spare 1/4 inch plate cutoff from a part I machined and turned it into a little crossbar to hold the parts in the anodizing bath
[17:30:18] <PetefromTn_> drilled and tapped a series of small holes across it to secure the TI wire to that will actually hold the parts and drilled and tapped another hole in the front to attach my positive wire to
[17:30:45] <PetefromTn_> mixed up the acid bath 50/50 to a total of 6gallons in the cooler about 2/3 full
[17:31:25] <PetefromTn_> mixed up the aluminum degreaser for the parts in another cooler with the DI water
[17:31:47] <PetefromTn_> prepared a large container of baking soda and water mix Just in case..
[17:32:19] <PetefromTn_> OOps this is linuxCNC devel...sorry LOL