#linuxcnc-devel | Logs for 2014-03-13

[00:01:24] <memleak> this channel goes OT!?
[00:01:55] <memleak> first time i've seen that!
[00:35:11] <seb_kuzminsky> memleak: don't tell cradek, he'll kick us out
[01:13:36] <memleak> says the one who just pinged him :P
[08:28:30] <alex_jon1> JT-Shop: around?
[08:28:34] <alex_jon1> jthornton: hi
[08:29:44] <jthornton> hi! for a minute
[08:29:53] <alex_jon1> good
[08:30:08] <jthornton> actually I'll see you out in the shop in a minute
[08:30:09] <alex_jon1> I changed some stuff for registrations, trying out now to see if I get a confirmation email
[08:30:27] <jthornton> Great!
[08:33:43] <JT-Shop> ok I'm out here now for a bit
[08:49:27] <alex_jon1> JT-Shop: lets do this in a /msg
[09:17:49] <alex_jon1> JT-Shop: a 'public' summary ;)
[09:18:03] <alex_jon1> there are multiple hosts that belong to DreamHost where linuxcnc.org is hosted
[09:18:21] <alex_jon1> when the website sends an email, a random outgoing server is selected
[09:18:43] <alex_jon1> some of the servers work, some don't as they are included in a DNS blacklist (like spamcop or dnsbl)
[09:19:04] <alex_jon1> here's an example: http://www.spamcop.net/w3m?action=checkblock&ip=
[09:24:46] <cradek> well that sucks
[09:26:04] <cradek> I bet big providers like dreamhost constantly have to fight this
[09:26:39] <archivist> with all the vulnerable websites that act as spam sources that users want to host
[09:26:54] <cradek> you need meticulous email hygeine to stay off all these various blacklists, and you simply can't have it with shared systems
[09:27:55] <cradek> sadly our only likely remedy is to stop depending on email
[09:28:16] <cradek> we (well swp) could complain to dreamhost but it can't possibly help very much
[09:28:31] <archivist> or change to a dedicated ip for our stuff
[09:28:51] <cradek> you mean move our hosting altogether? sure that would fix it
[09:29:03] <cradek> but jeez
[09:30:00] <archivist> or can we send via a clean host, not dreamhost
[09:30:37] <cradek> it's not clear to me how we could do that without moving the forum software
[09:31:10] <archivist> just change the settings
[09:31:22] <cradek> explain?
[09:31:32] <archivist> it should be able to use another host
[09:32:08] <cradek> oh you're assuming it uses smtp submission and you are suggesting to cause it to relay
[09:32:17] <cradek> ?
[09:32:35] <archivist> ie dont use dreamhost as the relay use gmail or something else
[09:33:29] <cradek> there's no such thing as a reputable open relay, so we'd need an arrangement (smtp auth) with a reputable relay
[09:33:44] <cradek> (assuming the forum uses smtp submission)
[09:33:52] <archivist> I agree it has to be a reputable relay
[09:35:42] <cradek> looks like it has smtp auth support
[09:38:04] <cradek> dreamhost may not allow outgoing smtp connections not going through their smarthosts
[09:38:08] <cradek> we'd have to try
[09:39:25] <cradek> [snotra]$ telnet mail.timeguy.com smtp
[09:39:25] <cradek> Trying
[09:39:25] <cradek> telnet: Unable to connect to remote host: Connection refused
[09:39:44] <cradek> so, nope (IMO this is a proper decision for them to make)
[09:42:12] <archivist> refused means your end doesnt it?, you didnt get no route to host
[09:42:31] <cradek> no, my end is fine
[09:42:42] <cradek> they're blocking it
[09:42:59] <cradek> with a packet filtering block you wouldn't expect to see no route to host
[09:43:56] <cradek> but maybe they allow submission
[09:44:33] <cradek> yeah they do allow outgoing submission
[09:44:44] <cradek> so it might be possible
[09:45:09] <cradek> brb
[10:02:50] <alex_joni> we can select php mail, smtp mail and another setting in the forum
[10:04:37] <cradek> personally I'd hate to relay all the forum's mail
[10:04:41] <cradek> I see at least two major problems:
[10:05:17] <cradek> 1, people report as spam things they don't want to see, even when they caused it (subscribed to threads accidentally, especially since this is our default, which I think is a questionable decision)
[10:05:40] <cradek> 2, forum subscription emails are one way to try sending out actual spam, or more likely, just harass a certain person
[10:06:03] <cradek> these kinds of emails do tend to get your mail server on blacklists, whether that's fair or not
[10:07:43] <cradek> alex_joni: can it send the regular forum emails (resulting from being subscribed to threads) to dreamhost, and the subscription emails to a separate smtp submission agent one of us controls?
[10:08:20] <cradek> (I'd guess no - those settings apply to all emails)
[10:08:29] <alex_joni> cradek: I think so
[10:08:40] <alex_joni> we actually care about registration emails now
[10:08:47] <alex_joni> and those are not sent by the forum email
[10:09:02] <alex_joni> but rather by the CMS (joomla + community builder extension)
[10:09:13] <alex_joni> so we can keep forum as is
[10:09:17] <cradek> interesting
[10:09:21] <cradek> that's a lot less bad then
[10:09:24] <alex_joni> and only change registration through an external smtp
[10:11:04] <cradek> all we need is some brave soul to run the relay
[10:20:21] <cradek> and silence falls
[10:23:12] * JT-Shop just got back from town and reads back
[10:23:51] <seb_kuzminsky> zultron: i think the 'parallel make of the docs sometimes fails' problem we talked about way back when is fixed now, 6e3b6a43f94
[10:25:36] <seb_kuzminsky> jepler: commit bee611dd8f isn't working for me - when i build the docs i still see all the .html files in docs/src
[10:26:25] <seb_kuzminsky> i think ** is a bash extension to globbing, it's not mentioned in glob(7) or glob(3) of fnmatch(3)
[10:26:40] <JT-Shop> what is an external smtp?
[10:29:04] <KGB-linuxcnc> 03Sebastian Kuzminsky 05v2.5_branch b44a252 06linuxcnc 10src/.gitignore ignore the new html docs stamp files * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=b44a252
[10:29:21] <cradek> seb_kuzminsky: from man gitignore: Two consecutive asterisks ("**") in patterns matched against full pathname may have special meaning:
[10:29:33] <seb_kuzminsky> the "may" sounds ominous
[10:29:44] <cradek> it's followed by many inscrutable bullet points
[10:29:59] <seb_kuzminsky> my gitignore(5) does not say that
[10:30:06] <seb_kuzminsky>, ubuntu precise
[10:30:12] <cradek> git version
[10:30:19] <seb_kuzminsky> aha
[10:30:40] <seb_kuzminsky> you frisky youngsters have newer better tools than us stodgy old conservative folk
[10:31:18] <cradek> ha
[10:31:20] <seb_kuzminsky> imma fix it for me
[10:31:39] * cradek pounds madly on his 1986 Model M
[10:36:37] <KGB-linuxcnc> 03Sebastian Kuzminsky 05master c2dcc7d 06linuxcnc 10.gitignore fix gitignore to ignore generated docs * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=c2dcc7d
[10:36:37] <KGB-linuxcnc> 03Sebastian Kuzminsky 05master 5aad39a 06linuxcnc 10src/.gitignore Merge remote-tracking branch 'origin/v2.5_branch' * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=5aad39a
[10:38:55] <cradek> yay
[10:45:46] <alex_joni> JT-Shop: SMTP = simple mail transfer protocol (e.g. email)
[10:46:06] <alex_joni> external smtp - is a smtp server that can send email, but is not hosted at dreamhost
[10:46:13] <JT-Shop> ah ok
[10:46:36] <JT-Shop> how do you get the email to the external smtp?
[10:47:04] <alex_joni> you tell joomla to send it using that external smtp (it gets generated there)
[10:47:32] <alex_joni> currently it's set up to generate the email on the same system where joomla is installed
[10:52:03] <cradek> JT-Shop: executive summary is our mail comes from machines that send all of dreamhosts's mail, and some of that is spammy, so that sending machine sometimes gets blacklisted
[10:52:48] <cradek> it's one of the curses you get when you use shared hosting
[10:53:25] <skunkworks> questions on the scratch build. If I add the scratch build repositories then do a sudo apt-get update then sudo apt-get install linuxcnc I get dgarrs build... if I install the circular-rc1 deb - it works - but if I do a sudo apt-get install linuxcnc (or do the update manager) it then re-installs the dgarrs deb.. (I know this is a flaw in the scratch builds.. but is there a work around?
[10:54:06] <alex_joni> you can specify the version of the package you want to install
[10:54:42] <cradek> yeah you're always going to get the "newest" one, where newest is according to debian package version sorting rules
[10:54:52] <skunkworks> I am trying to setup direction to use the debs for mere mortals... but if they run the update manager - it will install the latest deb
[10:54:53] <alex_joni> sudo apt-get install linuxcnc=mumble
[10:54:54] <cradek> you can pin to a certian version, which isn't really what you want either
[10:56:04] <skunkworks> bbl
[10:56:05] <cradek> well it's worse than that, because they're sorted by SHA
[10:56:28] <seb_kuzminsky> skunkworks: the scratch debs are deliberately made to sort as "older than" the regular debs
[10:56:49] <cradek> scratch isn't suitable for use as a regular deb "source"
[10:57:00] <seb_kuzminsky> agreed
[10:57:39] <cradek> we'd need scratch/branchname/package.increasing-numbers-from-a-tag
[10:59:02] <seb_kuzminsky> if debian supported wildcards in apt sources.list files we could use $BRANCH/increasing-numbers-from-a-tag
[10:59:22] <seb_kuzminsky> "deb http://buildbot.linuxcnc.org precise *"
[11:00:23] <cradek> if my grandmother had wheels...
[11:01:31] <seb_kuzminsky> oh the things we would do if your grandmother had wheels
[11:01:36] <alex_joni> heh
[11:01:55] <alex_joni> we could also tweak the package name? linuxcnc-branch
[11:02:00] <alex_joni> provides linuxcnc
[11:02:09] <alex_joni> so you can only install one at a time
[11:02:52] <cradek> install name=version doesn't pin the version number does it? it still gets upgraded next time
[11:03:02] <seb_kuzminsky> cradek: i think that's true
[11:03:03] <alex_joni> cradek: yes
[11:03:06] <seb_kuzminsky> alex_joni: that's an interesting idea
[11:04:12] <seb_kuzminsky> but i think it solves a problem that only skunkworks has
[11:04:20] <alex_joni> right
[11:04:29] <seb_kuzminsky> up until now no one but developers (or no one, maybe) cared about scratch debs
[11:04:38] <cradek> well the wider problem is you can't use normal aptness to get upgrades along a development branch
[11:05:04] <alex_joni> since git tags are not alphabetically sorted
[11:05:05] <cradek> (but I agree with you, this isn't a common problem)
[11:05:35] <seb_kuzminsky> if sources.list supproted wildcard components (like above), then we could have a component per branch
[11:05:54] <seb_kuzminsky> the buildslaves could find all the dsc files (because they would use wildcards in their apt)
[11:06:15] <seb_kuzminsky> users would not use wildcards, they'd select the component (aka branch) they wanted in their sources.list
[11:06:57] <seb_kuzminsky> user upgrades would work right, they would get newer debs from their branch and not accidentally switch to a different branch
[11:07:22] <cradek> maybe we could use target-release (install -t lucid-somebranch)
[11:08:02] <alex_joni> or apt-get install linuxcnc/targetbranch
[11:08:04] <seb_kuzminsky> the release is already part of the apt line, separate from the component
[11:08:25] <alex_joni> if you have more than one releases then you can select between them I think
[11:08:43] <seb_kuzminsky> bbl
[11:09:20] <cradek> I guess let's just answer "no" to skunkworks's question "is there a workaround?"
[11:10:03] <alex_joni> I'd answer "we think no, but prove us wrong" :)
[11:10:36] <cradek> I could argue it's a feature that upgrade puts you back to normal
[11:43:49] <linuxcnc-build> build #60 of 4017.deb-wheezy-armhf is complete: Failure [4failed apt-get-update shell_3] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/4017.deb-wheezy-armhf/builds/60 blamelist: Sebastian Kuzminsky <seb@highlab.com>
[11:53:28] <KGB-linuxcnc> 03Sebastian Kuzminsky 05unified-build-candidate-3 c742130 06linuxcnc 10(5 files in 5 dirs) Merge remote-tracking branch 'origin/master' into unified-build-candidate-3 * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=c742130
[12:14:14] <zultron> seb_kuzminsky, great! (make -j docs fix) I never figured out how to tickle that reliably.
[12:51:14] <skunkworks> well - it doesn't put you back to 'normal..' it puts you back to whatever the newest scratch build is..
[12:53:25] <skunkworks> or I a not understanding what is normal ;)
[13:36:01] <linuxcnc-build> build #61 of 4017.deb-wheezy-armhf is complete: Failure [4failed apt-get-update shell_3] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/4017.deb-wheezy-armhf/builds/61 blamelist: Sebastian Kuzminsky <seb@highlab.com>
[14:03:21] <linuxcnc-build> build #69 of 1401.rip-wheezy-amd64 is complete: Failure [4failed compile runtests] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1401.rip-wheezy-amd64/builds/69 blamelist: Norbert Schechner <nieson@web.de>, Sebastian Kuzminsky <seb@highlab.com>, Chris Radek <chris@timeguy.com>, Dewey Garrett <dgarrett@panix.com>
[14:05:58] <linuxcnc-build> build #68 of 1400.rip-wheezy-i386 is complete: Failure [4failed compile runtests] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1400.rip-wheezy-i386/builds/68 blamelist: Norbert Schechner <nieson@web.de>, Sebastian Kuzminsky <seb@highlab.com>, Chris Radek <chris@timeguy.com>, Dewey Garrett <dgarrett@panix.com>
[14:09:09] <skunkworks> wow - lets see how complicated we can make a rep rap?
[14:12:38] <linuxcnc-build> build #98 of 1403.rip-wheezy-armhf is complete: Failure [4failed apt-get-update compile runtests] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1403.rip-wheezy-armhf/builds/98 blamelist: Norbert Schechner <nieson@web.de>, Sebastian Kuzminsky <seb@highlab.com>, Chris Radek <chris@timeguy.com>, Dewey Garrett <dgarrett@panix.com>
[14:12:58] <cradek> skunkworks: I don't suggest adding the scratch directory as a source (in sources.list or equivalent). if you don't do that, you don't get any of the crazyness.
[14:13:22] <cradek> that's a dumping ground for debs built from unknown branches, not a deb repository that makes sense
[14:13:46] <skunkworks> oh - so if I don't add the sources I can do a sudo dkpg -i whatever.dev and it will install?
[14:13:53] <cradek> you can install debs from there with dpkg -i and not get any weirdness. when you run upgrade, you'll go back to whatever is in your sources
[14:13:57] <cradek> yes exactly
[14:14:17] <skunkworks> ok - how do make it install all the dependancies?
[14:14:47] <skunkworks> when I tried this I was missing a few things that it scolded me about.
[14:15:06] <cradek> after dpkg -i, you can do apt-get -f install and it will work out new dependencies if possible
[14:15:25] <skunkworks> (then adding sources and doing a apt-get updated fixed iirc
[14:15:37] <skunkworks> ah - ok
[14:15:40] <skunkworks> I will try that.
[14:16:14] * skunkworks goes to find a fresh install of linuxcnc...
[14:16:55] <skunkworks> they are getting harder and harder to find around here. I should have taken a snapshot of my vm before I mucked it up
[15:07:24] <linuxcnc-build> build #1915 of 0000.checkin is complete: Failure [4failed] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/0000.checkin/builds/1915 blamelist: Norbert Schechner <nieson@web.de>, Sebastian Kuzminsky <seb@highlab.com>, Chris Radek <chris@timeguy.com>, Dewey Garrett <dgarrett@panix.com>
[15:15:03] <cradek> x = joints[0]; ... left_virtual = atan(-y/x);
[15:20:41] <skunkworks> charles thing?
[15:39:18] <cradek> yeah
[15:39:26] <cradek> I tried to help him
[15:39:48] <cradek> I don't envy his deadline
[15:44:02] <cradek> viesturs also thinks it's a kinematics nightmare :-/
[15:44:30] <cradek> if I understand wallykins it looks like the normal working area is near a singularity
[15:45:16] <cradek> I bet the goal was to use only rotary joints which can be made more easily than slidey joints -- but I wonder why not use delta
[15:46:50] <cradek> https://github.com/NicholasSeward/ConceptFORGE/blob/master/Wally/DOCUMENTATION/I
[15:46:53] <cradek> oops
[15:47:07] <cradek> https://raw.github.com/NicholasSeward/ConceptFORGE/master/Wally/DOCUMENTATION/INVERSE%20KINEMATICS.png
[15:54:57] <cradek> I think the code is basic converted to python converted to C...
[15:55:48] <cradek> joints[0] = left_stepper*200/M_PI;
[15:56:12] <cradek> I think this is a conversion from radians to stepperdegrees
[15:58:58] <cradek> alex_joni: so you turned off the emails?
[16:01:40] <jthornton> cradek, I did
[16:02:34] <cradek> do you think that's an adequate long-term fix, or do we need the email process?
[16:03:59] <jthornton> let's let it run like this for a bit and see if we don't have any spammers
[16:04:34] <jthornton> I'm going to PR for 5 days so I can't manual confirm anyone till late next week
[16:07:54] <cradek> what's PR?
[16:10:18] <jthornton> Puerto Rico, where my wife is from... sadly we are going to scatter her mothers ashes into the sea at her birthplace
[16:10:43] <cradek> heck, sorry to hear that
[17:18:24] <KGB-linuxcnc> 03Sebastian Kuzminsky 05unified-build-candidate-3 1423631 06linuxcnc 10debian/configure posix build-depends on libudev-dev * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=1423631
[17:43:08] <andypugh> Well, it turns out that the guy asking about toolchangers has decided to go with Mach
[18:06:43] <Tom_itx> he must have given up
[18:07:03] <andypugh> Yeah :-(
[18:07:26] <Tom_itx> i don't recall him asking very much before he did either
[18:08:09] <andypugh> He wanted to pay someone to set up LinuxCNC for him, in Florida. But nobody was interested.
[18:09:35] <Tom_itx> not a self starter ehh?
[18:11:40] <andypugh> He wanted the machine productive in 2 weeks. Nobody in LinuxCNC-land thought that was possible. I have notied that Mach users are much more optimistic in that regard. And much less accurate in their optimism.
[18:12:31] <Tom_itx> not sure what he needed done but that sounds like a short timeframe to production
[18:13:45] <andypugh> There is a toolchanger, that was the sticking point.
[18:28:04] <cradek> andypugh: was it you who advised he get the drawbar wired up to a pushbutton and get his parts made so the pinch is over? that sounded like good advice to me.
[18:29:04] <andypugh> "Use LinuxCNC not Mach" was also probably ggod advice, but he ignored it.
[18:29:42] <linuxcnc-build> build #46 of 4012.deb-precise-rtpreempt-i386 is complete: Failure [4failed apt-get-update shell_2] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/4012.deb-precise-rtpreempt-i386/builds/46 blamelist: Sebastian Kuzminsky <seb@highlab.com>
[18:30:06] <linuxcnc-build> build #46 of 4013.deb-precise-rtpreempt-amd64 is complete: Failure [4failed apt-get-update shell_2] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/4013.deb-precise-rtpreempt-amd64/builds/46 blamelist: Sebastian Kuzminsky <seb@highlab.com>
[18:32:12] <KGB-linuxcnc> 03Dewey Garrett 05dgarr/demo_appsv4 e5cd156 06linuxcnc 10(10 files in 6 dirs) pickconfig: copy demo dirs only if allowdemocopy * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=e5cd156
[18:47:10] <linuxcnc-build> build #47 of 4011.deb-precise-xenomai-amd64 is complete: Failure [4failed apt-get-update shell_3] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/4011.deb-precise-xenomai-amd64/builds/47 blamelist: Sebastian Kuzminsky <seb@highlab.com>
[19:30:42] <linuxcnc-build> build #48 of 4010.deb-precise-xenomai-i386 is complete: Failure [4failed apt-get-update shell_3] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/4010.deb-precise-xenomai-i386/builds/48 blamelist: Sebastian Kuzminsky <seb@highlab.com>
[19:35:23] <linuxcnc-build> build #1480 of 4006.deb-lucid-rtai-i386 is complete: Failure [4failed shell_3] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/4006.deb-lucid-rtai-i386/builds/1480 blamelist: Sebastian Kuzminsky <seb@highlab.com>
[20:27:19] <linuxcnc-build> build #63 of 4017.deb-wheezy-armhf is complete: Failure [4failed apt-get-update shell_3] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/4017.deb-wheezy-armhf/builds/63 blamelist: Dewey Garrett <dgarrett@panix.com>
[21:24:43] <skunkworks> http://www.cnczone.com/forums/linuxcnc-formerly-emc2/206712-new-trajectory-planner-testers-programs-wanted-post1454602.html#post1454602
[21:24:58] <skunkworks> that is with the current tp.. Go figure.
[21:25:16] <skunkworks> I have no clue what he was seeing before..