#emc-devel | Logs for 2006-03-26

[00:02:15] <jmkasunich> got an ubuntu install question
[00:02:18] <jmkasunich> partitioning
[00:02:22] <jmkasunich> I have 2x18G disks
[00:02:35] <jmkasunich> I'm gonna want to have lots of qemu virtual machines
[00:03:03] <jmkasunich> where do they put their "disks"? in /? in /usr? in /home?
[00:03:42] <jmkasunich> can I use lvm or something to make the two disks work like one big one, or should I put / on one disk and /home on another?
[00:05:31] <alex_joni> jmkasunich: you said you want to restore /home at some point? which you had on another partition?
[00:06:11] <jmkasunich> not this box
[00:06:37] <jmkasunich> I'm installing ubuntu on a dual CPU server box, gonna try qemu to run the entire compile farm on one machine
[00:08:44] <alex_joni> ahh.. ok, then I guess lvm is ok
[00:08:49] <alex_joni> although I never tried it :D
[00:23:16] <cradek> jmkasunich: lvm works but the install is a little obtuse, it took me two tries
[00:23:31] <cradek> jmkasunich: when you are asked to name the volume I think you have to name it Ubuntu
[00:24:49] <jmkasunich> I'm googling, and it does sound obtuse
[00:26:13] <jmkasunich> it looks like it created two LVs, both on one disk
[00:26:20] <jmkasunich> the other disk seems unused
[00:26:39] <jmkasunich> one LV is swap, the other is /
[00:32:05] <jmkasunich> 255MB seems pretty big for a boot partition
[00:33:03] <jmkasunich> I have about 5 kernels on this box, and the total in /boot is 22M
[00:33:13] <jmkasunich> I think I'll set it to 40M
[00:43:04] <jmkasunich> cradek: what do you remember about setting up the LVM?
[00:43:32] <jmkasunich> the LVM howto is of limited usefulness - it assumes you have a working system and a command line, the installer interface is different
[00:47:08] <jepler> jmkasunich: maybe you should make a "/" of 2 or 3 gigs, then create the LVM after you reboot
[00:47:36] <jepler> jmkasunich: a 'server' install is well under 1 gig on disk, and it still is after you add X and all the development tools needed to rebuild emc and axis
[00:48:45] <jmkasunich> if / is a physical partition, then doesn't that mean that later LVM partitions can only be used for things below / like /home and /usr?
[00:48:56] <jepler> yes, that would be true
[00:49:36] <jmkasunich> what I want to do is have a physical /boot of 40M (no problem doing that), and have the remaining 18.16G on one disk and all of the 18.20G on the other part of one big logical volume
[00:49:53] <jmkasunich> then divide that big LV into swap and /
[00:49:56] <jmkasunich> and thats it
[00:51:26] <jmkasunich> the initial config it gave me was one drive only, split into a 255M physical /boot, and the remaining 17.6G as a LVM partition
[00:51:47] <jmkasunich> that 17.6G VG was split into two LV, / and swap
[00:52:23] <jmkasunich> I've since resized /boot to only 40M, delete the phys partition that was 17.6, and created a new LVM partition that is now 18.16G
[00:52:36] <jmkasunich> also created a LVM partition on the other disk of 18.2G
[00:52:54] <jmkasunich> but the VG still totals 17.6G and I can't seem to do anything with it
[00:53:11] <jepler> I'm not sure what to try next
[00:53:16] <jepler> I'm not sure cradek would remember either
[00:53:31] <jepler> we tried a bunch of stuff, including getting one install that wouldn't find the LVM partitions at the first boot..
[00:54:34] <jmkasunich> I just read somethign that recommended keeping / on a physical partition
[00:55:04] <jmkasunich> do you know where qemu sets up the files that is uses as "disks" for the virtual machines?
[00:55:42] <jepler> you create them with the 'qemu-img' command whereever you like
[00:56:00] <jepler> e.g., qemu-img create bdi4.hda 4G
[00:56:19] <cradek> wooo
[00:56:19] <jmkasunich> so I could do a 3G /, and later make /qemu as a separate partition will all the remaining disk space?
[00:56:24] <cradek> I got a picture
[00:56:28] <jepler> jmkasunich: yes
[00:56:29] <jmkasunich> yay!
[00:56:35] <cradek> with .5 ohm there
[00:56:59] <jmkasunich> 47*0.01 = yellow, purple, gold
[00:57:20] <cradek> sure could be that
[00:57:35] <cradek> all I have is .51 but it gives a nicely filled out picture
[00:57:42] <jmkasunich> close enough
[00:57:59] <cradek> the real test is whether it will still be working in an hour or week
[00:58:05] <jmkasunich> yeah
[00:58:18] <jmkasunich> don't leave it running unattended ;-)
[00:58:27] <jmkasunich> ok, disks...
[00:59:16] <jmkasunich> "qemu-img create bdi4.hda 4G" creates a 4G "disk" as a file in the current working dir?
[00:59:21] <jepler> right
[00:59:47] <cradek> read the qemu-img manpage first - there's a disk type that only uses the necessary space, you want that if possible
[01:00:16] <cradek> oh it's the default
[01:00:25] <jmkasunich> so for best flexibility I want a big honking /home, and I create the virtual disks there, as well as downloading iso's there, etc
[01:00:55] <jmkasunich> which brings me back to LVM
[01:00:57] <jepler> the other thing you might want is "qcow" images.
[01:01:11] <jmkasunich> the only way to make /home bigger than one disk is to use LVM, right?
[01:01:28] <jepler> A "qcow" can record only the changes compared to some other disk image (given by an absolute path)
[01:01:34] <SWPadnos> you can put the images anywhere you want - just put a few on the second disk
[01:01:51] <SWPadnos> the image to use is a command-line argument to qemu
[01:02:05] <cradek> jmkasunich: you might want to give yourself plenty of swap
[01:02:19] <jmkasunich> ignoring the qemu for a moment... I'm just doing the install of the base system right now
[01:02:26] <cradek> who knows how many VMs you'll end up with - and swap is free
[01:02:30] <jmkasunich> cradek: I was planning on it - at least a gig, maybe a couple
[01:02:47] <SWPadnos> how much memory is in the machine?
[01:02:51] <jmkasunich> 512M
[01:02:58] <SWPadnos> ok
[01:03:02] <jmkasunich> the default swap was 768M
[01:03:53] <jmkasunich> my problem is that I can't get the install to do anything usefull with the second disk
[01:04:45] <cradek> jmkasunich: I wish I could remember what we did, but I don't
[01:04:54] <cradek> I only remember it was hard to figure out
[01:05:27] <jmkasunich> what if I don't do LVM at all for the install?
[01:05:42] <jmkasunich> make a 40M /boot and a 3G /
[01:05:57] <jmkasunich> but the install will make a /home inside the 3G /
[01:06:05] <cradek> that's easy to fix afterward
[01:06:09] <cradek> just mount your new /home on top of it
[01:06:28] <jmkasunich> that sounds like a plan
[01:06:40] <jmkasunich> but I'll probably have to ask for help when I get to that point
[01:06:53] <jmkasunich> at least with a command line the LVM howto will be usable
[01:07:46] <SWPadnos> I wonder if you could create the LVM volume (spanning both disks) on another box, then have the install notice that it's meant to be a spanned volume?
[01:08:00] <jmkasunich> what other box?
[01:08:12] <SWPadnos> any otheer machine that has a running install and SCSI
[01:08:20] <jmkasunich> don't have one
[01:08:21] <SWPadnos> installation, that is
[01:08:23] <SWPadnos> oh
[01:08:26] <cradek> I know you can do it at the install step
[01:08:39] <jmkasunich> you just don't remember how
[01:08:48] <jmkasunich> and I can't figure it out either
[01:09:03] <cradek> well I do - you struggle and curse for a while, then try again, then it works
[01:09:04] <jmkasunich> I really don't need a big /
[01:09:07] <cradek> that's true
[01:09:11] <cradek> I like your new plan better anyway
[01:09:22] <jmkasunich> just trying to decide how big to make /
[01:09:34] <SWPadnos> 2^32
[01:09:52] <jmkasunich> really, the ONLY thing I want to run on the base box is qemu, and minimal support tools
[01:09:55] <cradek> my very feature-complete system uses 2.5G for everything
[01:09:55] <SWPadnos> (it's as good a number as any ;) )
[01:09:57] <cradek> including gnome
[01:10:08] <cradek> Filesystem 1K-blocks Used Available Use% Mounted on
[01:10:08] <cradek> /dev/sda1 8356868 2535724 5396628 32% /
[01:10:29] <jmkasunich> /dev/hdb1 5905972 2525260 3080700 46% /
[01:10:34] <jmkasunich> pretty close
[01:10:46] <cradek> so 4G is **plenty**
[01:10:49] <SWPadnos> Filesystem Size Used Avail Use% Mounted on
[01:10:50] <jmkasunich> but that includes my /home
[01:10:52] <SWPadnos> /dev/hda2 18G 2.4G 15G 14% /
[01:10:58] <cradek> 3G is probably enough especially without gnome
[01:11:26] <SWPadnos> that one is a full BDI 4.30 with kdevelop and friends installed
[01:11:38] <SWPadnos> and about 17 emc CVS checkouts
[01:16:12] <jmkasunich> if my / is a physical partition, do I need a separate /boot partition? I don't think so
[01:17:50] <jepler> I think you can do without it
[01:18:14] <jmkasunich> ok, I have a 3G / and a 1.5G swap, the rest is free space to be partitioned later
[01:18:17] <cradek> I think scsi has never really needed that
[01:19:22] <jmkasunich> I'm afraid....
[01:20:03] <jmkasunich> even tho I restarted the partitioning page, I never got to the screen that says "erase a whole disk and use it" as one of several options (where another was "erase a disk and use LVM")
[01:20:14] <jmkasunich> I still have some LVM entries showing up
[01:28:00] <jepler> cradek: any idea why the two blends are different?
[01:28:01] <jepler> http://emergent.unpy.net/index.cgi-files/sandbox/twoblends.png
[01:28:07] <jepler> both were made with the same 300% feed override
[01:28:20] <jepler> the top was made left-to-right, and the bottom was made right-to-left; maybe that's important?
[01:29:03] <jmkasunich> another qemu question? do I need a big swap on the base machine? or do I define a moderate swap for each virtual machine?
[01:29:51] <jepler> jmkasunich: I've given medium amounts of ram (e.g., 300 megs to 500 megs) and let the installers select an amount of swap
[01:29:57] <jepler> I don't know if that's a good idea or a bad one
[01:31:07] <jmkasunich> I'm partitioning for the base machine now - if each qemu install is going to make its own swap partition, I don't think I need a huge base swap partition do I?
[01:31:32] <SWPadnos> how do you plan to run the VMs?
[01:31:37] <jmkasunich> ?
[01:31:48] <SWPadnos> ie, always running, or somehow started and stopped whenever a recompile is needed
[01:31:56] <jmkasunich> always running I think
[01:31:59] <jepler> you need enough ram+swap for all the VMs that will be running at once
[01:32:07] <SWPadnos> right
[01:32:08] <jepler> so if that's 4 VMs at 500 megs each -> 2 gigs ram + swap
[01:32:08] <jmkasunich> oh
[01:32:29] <jmkasunich> I see, when you define the VM, you tell qemu how much ram that virtual box has
[01:32:31] <SWPadnos> the swap for the VMs would be on their virtual disks though
[01:32:35] <SWPadnos> yes
[01:32:39] <jepler> qemu -m sets the memory
[01:33:15] <jmkasunich> seems like best performance would be if I give each VM a lot of ram, and have a big base swapfile
[01:33:33] <jmkasunich> because if only one VM is active, it uses a lot of the physical ram
[01:33:59] <jmkasunich> if a VM has little ram and a big virtual swap, it will swap a lot even if the rest of the physical ram is empty
[01:34:05] <SWPadnos> all VMs will be running RT linux of some flavor - "active" may be a relative terms
[01:34:24] <jmkasunich> I won't actually be running RT code, just compiling for the most part
[01:34:30] <jepler> you can add additional swap too
[01:34:38] <jmkasunich> and running RT code actually uses very little RAM
[01:34:54] <jmkasunich> measured in 10s of K, not hundreds of M
[01:34:54] <SWPadnos> but it uses a lot of interrupts, and is just about never inactive
[01:35:27] <SWPadnos> though under normal circumstances, the VMs should still be calling idle() pretty often
[01:35:28] <jepler> if they're not booting desktops, these hosted systems will never touch their own swap
[01:35:48] <jepler> not with 300 megs emulated RAM
[01:35:59] <jmkasunich> so I can declare say 300M emulated RAM and 300M swap for each VM
[01:36:13] <SWPadnos> doesn't qemu allocat the entire emulated RAM amount at startup though?
[01:36:16] <jmkasunich> and make sure my base swap = 300M * number of VMs
[01:36:32] <SWPadnos> the swap is internal to the VM, it'll be on the virtual disk
[01:37:13] <SWPadnos> yes, I'd say base swap should be at least (#VMs * 300M) + extra for base system
[01:37:33] <jmkasunich> what happens if you have a 512M box, and create a VM with 1000M ram and 100M swap
[01:37:47] <jmkasunich> that VM has a total "memory" available of 1,1G
[01:37:47] <jepler> you may be unhappy with the performance
[01:38:02] <SWPadnos> the swap in the VM is immaterial to the base system RAM / swap config
[01:38:22] <SWPadnos> it's emulated physical RAM, and virtual disk space for emulated swap
[01:38:39] <jmkasunich> two options: VM with 1000M ram and 100M swap, vs VM with 300M ram and 800M swap
[01:38:46] <jmkasunich> one will swap to virtual disk
[01:38:54] <jmkasunich> the other will swap to the base swapfile
[01:38:57] <jmkasunich> which is better?
[01:39:11] <SWPadnos> err - good question ;)
[01:40:47] <jmkasunich> if the working set is <300M, probably no difference
[01:40:51] <jepler> I think both will do OK when the working set fits in host RAM (but 1000M RAM guest will be better)
[01:41:05] <jmkasunich> if its >512M< maybe no differnce (both will swap)
[01:41:09] <jepler> (if the workng set is from 300 to 1000M)
[01:41:36] <jmkasunich> if the working set is between 300 and 500, the 1000M ram setup will be better, because the 300M ram VM would start swapping
[01:42:10] <jmkasunich> fsck it
[01:42:36] <jmkasunich> I'll just declare 300M * about 6 + 700 for the base = 2.5G swap for the base system
[01:44:18] <jmkasunich> 3G / on sda, 3G swap on sdb, the remaining 15.2G on each disk will be merged later into a LVM /home
[01:45:49] <jmkasunich> gonna have to leave in 15 mins or so anyway... gawd I wasted a lot of time trying to decide what to do :-(
[01:46:17] <jmkasunich> "analysis paralisis"
[02:35:10] <cradek> whee
[02:35:23] <SWPadnos> ok, what did you fix now? ;)
[02:41:20] <cradek> my monitor
[02:42:41] <cradek> it was a throwaway (dead) apple monitor but it's a multisync that does 1600x1200 at 75Hz, which my old one couldn't do
[02:43:08] <cradek> total parts required: one transistor, one resistor
[02:51:19] <cradek> and of course 6? hours of my life
[03:02:16] <SWPadnos> heh, good deal
[03:02:52] <cradek> the huge 1500v output transistor cost $25 - crazy
[03:03:04] <SWPadnos> that's an expensive transistor
[03:03:10] <SWPadnos> but a cheap monitor, overall
[03:03:26] <cradek> yep, glad I only had to buy one... often you go through several before you find all the problems
[03:03:57] <SWPadnos> yep, that's not a good diagnostic throwaway part
[03:04:21] <cradek> I think this monitor would have hit the recycle pile before I would have bought another one though
[03:04:41] <SWPadnos> how old is the monitor?
[03:04:41] <cradek> so it's all good, I like it when I can keep something from being thrown away and get something nice out of it too
[03:04:56] <SWPadnos> yeah, a double win
[03:05:06] <cradek> 97 I think
[03:05:25] <SWPadnos> oh, an old, but very high end, apple monitor
[03:05:26] <cradek> it's pretty bright - might not have a lot of hours on it (has been broken for a while maybe)
[03:06:18] <SWPadnos> I finally broke down and bought one of these:
[03:06:22] <SWPadnos> http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=8786272728
[03:06:57] <cradek> holy cow
[03:07:06] <SWPadnos> it's very high res ;)
[03:07:23] <cradek> like jeff's laptop I bet the internet is about two inches wide
[03:07:33] <cradek> (and too small to read)
[03:07:38] <SWPadnos> yep, google earth fits on one screen
[03:07:53] <cradek> is that a tube or lcd?
[03:07:57] <SWPadnos> I saw one of these at a trade show in Germany a couple of years ago
[03:07:59] <SWPadnos> LCD
[03:08:11] <SWPadnos> it's bright, and the images look like photographs
[03:08:38] <cradek> heh, I remember when megapixel displays were the big thing - yours is almost 10
[03:08:39] <SWPadnos> the pixel resolution is the same as a super-fine fax, but in full color
[03:08:43] <SWPadnos> yep
[03:08:56] <cradek> 200dpi? wow
[03:09:00] <SWPadnos> yep, 204
[03:09:03] <jepler> that's quite the monitor
[03:09:03] <cradek> my lcd is 100
[03:09:29] <jepler> I thnk my laptop is 150
[03:09:39] <SWPadnos> I've seen some at 133 or thereabouts
[03:09:39] <jepler> hypot(1920,1200)/15.1 = 149.9
[03:09:55] <cradek> SWPadnos: what's retail on that thing?
[03:10:08] <SWPadnos> it was &7500 for the Viewsonic
[03:10:12] <SWPadnos> $7500
[03:10:18] <SWPadnos> the IBM version was $9k
[03:10:23] <cradek> holy effing crap
[03:10:25] <jepler> http://froogle.google.com/froogle?q=viewsonic+3840x2400&hl=en&hs=wmf&lr=&client=firefox-a&rls=org.mozilla:en-US:official&sa=N&tab=ff&sa=X&oi=froogle&ct=title
[03:10:41] <jepler> that's a bit out of my price range
[03:10:49] <cradek> that's 300x what this monitor cost :-)
[03:10:52] <SWPadnos> yeahoh right - I think I saw Tiger Direct or someone at the $5k mark
[03:10:55] <SWPadnos> heh
[03:11:25] <cradek> 300x the price for 5x the pixels? hmmmmm
[03:11:45] <SWPadnos> well, I only spend like 50x what you did ;)
[03:11:48] <SWPadnos> spent
[03:12:05] <cradek> ah, you got a sweet deal then
[03:12:06] <SWPadnos> but no time. if you count the time, then it's only 3x, so a better deal per pixel
[03:12:16] <cradek> true
[03:12:26] <cradek> does that somehow take two dvi ports to drive?
[03:12:37] <SWPadnos> yep. two dual-link ports
[03:12:47] <jepler> so another $300 or more for the video card?
[03:12:48] <SWPadnos> it can run on a single port, but the refresh rate is slower
[03:13:11] <SWPadnos> I have a Radeon that can drive it, plus a Matrox Parhelia, and a 7800GT that may do it
[03:13:34] <SWPadnos> then again, trying to wrangle 9MP would bring any of those cards to a crawl
[03:13:36] <cradek> parhelia sounds like some kind of disorder to me
[03:13:46] <SWPadnos> heh - gamers think it is ;)
[03:14:14] <SWPadnos> it's actually the bright spots to the sides of a solar eclipse
[03:14:30] <SWPadnos> or a Matrox video card that handles 3 monitors
[03:14:36] <cradek> http://en.wikipedia.org/wiki/Paraphilia
[03:14:54] <SWPadnos> that's not quite the same thing ;)
[03:15:08] <SWPadnos> http://en.wikipedia.org/wiki/Sun_dog
[03:15:34] <cradek> cool, I learned a word
[03:16:07] <SWPadnos> always room to learn new words
[03:16:16] <cradek> yep, brain's not full yet
[03:16:21] <jepler> I thought maybe it was this word misspelled: http://www.answers.com/perihelion
[03:16:22] <SWPadnos> there are over a million of them (in English), and most people only use about 4500 or so
[03:16:35] <SWPadnos> the opposite of aphelion
[03:17:11] <cradek> jepler: have you been able to get firefox in ubuntu to let you paste a link into the window to go to it?
[03:17:29] <cradek> I thought it was an option, but I don't see it (and it's not on by default)
[03:17:43] <SWPadnos> can you drag text onto the empty part of the tab bar?
[03:17:52] <SWPadnos> (text = a URL)
[03:18:05] <jepler> cradek: you mean select text in rxvt, then middle-click in firefox? That worked for me, but I've manually installed firefox 1.5.
[03:18:13] <jepler> SWPadnos: drag? text? You must be kidding.
[03:18:14] <cradek> hmm, ok
[03:18:16] <SWPadnos> heh
[03:18:21] <SWPadnos> XDnD?
[03:18:32] <jepler> that's the thing you always do accidentally when you use one of those godawful broken text editors where all the letters have different widths.
[03:19:00] <SWPadnos> anti-proportional fonts, are we? :)
[03:19:10] <jepler> they're a fad
[03:19:22] <SWPadnos> heh
[03:19:36] <SWPadnos> I think they've been around longer than monospace, by a few thousand years
[03:19:57] <cradek> I thought colors in a text editor were a fad too, but I've grown to appreciate it
[03:20:07] <cradek> give me another decade or two and maybe I'll accept proportional fonts
[03:20:14] <SWPadnos> I almost can't read code without syntax hilighting any more
[03:20:24] <SWPadnos> proportional for code is impossible though
[03:21:06] <cradek> if I had to pick one, I'd rather have autoindent than syntax highlighting
[03:21:32] <cradek> especially for lisp and python, less important for C
[03:21:40] <SWPadnos> working autoindent is good, but I'm so used to broken editors that it doesn't matter much
[03:21:53] <cradek> just say no
[03:22:31] <SWPadnos> I do. I tried to use Kate, and I think I had it right for a few minutes once, but I'm not sure
[03:22:40] <SWPadnos> I use it, but without autoindent
[03:22:59] <cradek> I've never used kate (or kde) but it seems like a lot of people like it
[03:23:28] <SWPadnos> it's pretty good. lots of languages for syntax hilighting, and good multi-window support
[03:23:41] <SWPadnos> though there's no "synchronized scrolling" that I've seen
[03:23:49] <cradek> what's that?
[03:24:18] <cradek> does it have good support for tags? It seems like gui editors have forgotten how to do that
[03:24:20] <SWPadnos> line up a split window on two files (or two parts of the same file), and the scrollbar moves them in unison
[03:24:31] <SWPadnos> tags like emacs>
[03:24:33] <SWPadnos> ?
[03:24:40] <cradek> yeah, like vi and emacs
[03:24:46] <SWPadnos> I don't think so
[03:24:48] <cradek> jump to the function whose call is under the cursor
[03:24:53] <cradek> now, jump back to where I was
[03:25:15] <cradek> I'd be really unhappy without that
[03:25:22] <SWPadnos> sort of. it can collapse functions, loops, and comments
[03:25:40] <SWPadnos> there are probably keystrokes to go to matching parens and the like
[03:25:47] <cradek> that's not really what I mean at all
[03:25:54] <SWPadnos> but nothing like "go to header file where this function is declared"
[03:26:16] <cradek> say there's a call fun(a,b,c) the editor should know how to take me to fun()
[03:26:16] <SWPadnos> there is a ctags-like plug-in. I may enable it one day
[03:26:23] <cradek> ah
[03:26:27] <SWPadnos> right, and back to where you jumped from
[03:26:44] <cradek> yeah, and infinite jumps deep
[03:26:53] <cradek> and defines, etc.
[03:27:30] <SWPadnos> well, it is extensible, but I have no idea what the mechanism is
[03:28:36] <SWPadnos> hmm - there is a modeline plugin, that "sets document settings based on vim or emacs modelines"
[03:30:58] <cradek> hmm, I had dinner but I'm hungry again
[03:31:01] <cradek> see you later
[03:31:08] <SWPadnos> heh. see you later
[03:31:55] <jepler> In a folder with 392 messages I've sent, I used 6808 distinct words, 3515 of them more than once.
[03:32:03] <SWPadnos> heh
[03:32:22] <SWPadnos> is that the message bodies only, or does it include names and addresses (and technical terms)?
[03:32:26] <jepler> bodies
[03:32:48] <jepler> I processed the folder with this program: http://emergent.unpy.net/index.cgi-files/sandbox/vocab.py
[03:33:22] <jepler> "the to a axis" are my 4 most frequent words
[03:33:23] <SWPadnos> I wonder if that would work on Windows Python
[03:33:39] <SWPadnos> axis - who would have thought
[03:33:40] <jepler> It is probably portable, but you have to give it a unix "mbox" style mailbox
[03:33:55] <SWPadnos> I thikn mozilla uses that
[03:33:57] <SWPadnos> think
[03:34:10] <jepler> probably needs python2.4 too
[03:34:23] <SWPadnos> I believe that's what I've got
[03:36:12] <SWPadnos> hmmm - maybe I don't have python on this machine
[03:36:34] <jepler> *astonished gasp*
[03:36:55] <jepler> I just couldn't do without Python
[03:37:00] <jepler> or something very much like it
[03:37:17] <SWPadnos> hmmm - 2.2 only
[03:37:25] <SWPadnos> I winder if that was the cygwin installed version
[03:39:12] <SWPadnos> have you tried the latest version (2.4.3c1)?
[03:39:23] <jepler> no
[03:39:37] <jepler> looks like 2.4.2 is the version ubuntu installed
[03:39:49] <SWPadnos> ah - latest production version is 2.4.2
[03:40:01] <SWPadnos> 2.4.3c1 was released March 23, 2006
[03:40:43] <jepler> I'm sure it's got some bugs fixed
[03:40:51] <SWPadnos> at least 50
[03:43:43] <jepler> oh darn, this was one of my favorite stupid python bugs: SF Bug #887946: fix segfault when redirecting stdin from a directory.
[03:43:46] <jepler> $ python < /
[03:43:48] <jepler> Segmentation fault.
[03:43:58] <SWPadnos> bummer
[03:44:05] <SWPadnos> was that not fixed?
[03:44:19] <jepler> the release notes say it is
[03:44:25] <SWPadnos> cool
[03:44:29] <jepler> it's the only python bug I knew offhand how to provoke
[03:44:35] <SWPadnos> heh
[04:01:51] <skunkworks> I almost picked up a python book at barns and noble. I thought I need to read my "debian gnu/linux 3.1 bible" and "moving to linux" second adition first
[04:02:22] <jepler> I bought a python book once. never used it.
[04:03:11] <cradek> I don't remember offhand the last technical book I bought
[04:03:16] <cradek> I rarely buy them
[04:03:23] <skunkworks> the "moving to linux" book has a subtitle of "kiss the blue screen of death goodbye" ;)
[04:03:52] <skunkworks> good point. I don't think I have bought a book for work in ages.
[04:04:50] <cradek> I bet it was (sadly) the sendmail book
[04:04:59] <cradek> it was a long time ago
[04:05:13] <skunkworks> I don't really have any base for linux yet
[04:05:36] <jepler> over the last year I've bought a couple of Nutshell books.
[04:05:46] <jepler> It taught me that you can't cover C++ in two nutshell books
[04:05:55] <cradek> skunkworks: you're a new unixer?
[04:06:09] <skunkworks> very new
[04:06:30] <skunkworks> emc did it.
[04:08:03] <cradek> hope you find that linux meets your other needs too - it sure does for me
[04:08:49] <skunkworks> I could get used to it. - looking more towards mysql for work.
[04:10:03] <cradek> yeah mysql seems fine but I'm not a power user or anything
[04:11:36] <skunkworks> I have a ton of stuff I have written in access/vba that is getting too big. looking for the next development software (yes access)
[04:11:39] <SWPadnos> jepler, are you supposed to run vocab.py < my.mbox ?
[04:12:07] <SWPadnos> eeewww - access ;)
[04:12:50] <cradek> skunkworks: access is a toy isn't it? phpbb says if you intend to have more than one user at a time, don't use access because it'll get all confused
[04:13:50] <jepler> SWPadnos: yeah, I think it reads stdin
[04:13:51] <jepler> I already forgot
[04:13:52] <skunkworks> that is what I hear. I must be lucky because I have over 30 users into one database. suprisingly fast.
[04:13:54] <SWPadnos> heh
[04:13:56] <jepler> and it'll spew a bunch of stuff to stdout
[04:14:11] <cradek> skunkworks: well if you have to learn something new anyway, it's good to pick free software
[04:14:22] <skunkworks> I hired a programmer that can't believe how fast my databases run.
[04:14:22] <skunkworks> right
[04:14:33] <cradek> maybe you're lucky or you know something they don't...
[04:14:39] <skunkworks> we are cheap
[04:14:47] <jepler> Silicone lubes
[04:14:47] <jepler> If you are looking for a lube that won't dry out for a long time, try Eros. Silicone lubes feel like oil, but are latex compatible. They stay slick in water (hot tub, anyone?) and are great for anal sex. Silicone lube doesn't wash away unless you use soap and water. Don't use silicone lubes on silicone or cyberskin dildos, unless you want to see your dildo melt--cover such toys with a condom before you use them with a silicone-based lube.
[04:15:03] <skunkworks> the only thing he can think of is that it is running on an old 4.11 novell server ;)
[04:15:10] <cradek> jepler: ?
[04:15:12] <jepler> (that piece of advice is free)
[04:15:44] <skunkworks> http://www.electronicsam.com/images/TUB.JPG ?
[04:15:45] <cradek> well, that's good to know, but a bit off-topic
[04:15:52] <jepler> I'd rather talk about anything than Microsoft Access
[04:16:07] <cradek> I understand
[04:16:17] <skunkworks> so do I actually
[04:17:07] <jepler> hmm .. if only google would tell you the pages least related to any given search terms
[04:18:55] <cradek> what page is least related to microsoft access?
[04:19:18] <cradek> yay, got my state refund already
[04:19:38] <jepler> I've gotten my federal, but not state
[04:23:39] <skunkworks> Taxes?
[04:23:42] <skunkworks> ;)
[04:25:20] <SWPadnos> jepler, ok, it ran fine on my sent folder, the count is 9238 4867
[04:25:25] <SWPadnos> what's the second number?
[04:25:37] <SWPadnos> (oh, it's a 20M folder, with ~1300 messages)
[04:26:17] <SWPadnos> ah, the number used more than once
[04:29:23] <cradek> 9022 4811
[04:29:51] <SWPadnos> well, I guess we're all above average ;)
[04:29:53] <cradek> hmm, the most common word I used is "g"
[04:29:57] <SWPadnos> heh
[04:30:47] <SWPadnos> "the to a i and you" are my top 6
[04:31:47] <SWPadnos> though the output may not be accurate on my machine, due to the file format. I see xmozillastatus in there with 623 occurrences
[04:32:01] <SWPadnos> and I'm pretty sure this is the first time I've typed that word
[04:32:14] <cradek> there was a little crap in my output too, but not much
[04:32:48] <SWPadnos> only 3 misspellings of the as hte though - that's good
[06:07:10] <SWPadnos_> SWPadnos_ is now known as SWPadnos
[07:09:50] <SWPadnos_> SWPadnos_ is now known as SWP_Away
[10:34:31] <alex_joni> morning everyone
[15:32:14] <dan_falck> rayh: good morning
[15:37:27] <jmkasunich> morning folks
[15:37:44] <rayh> Hi Dan.
[15:37:55] <rayh> Hi jmk
[15:38:24] <dan_falck> rayh: I can get into my Networksolutions account
[15:38:45] <rayh> Hey jepler Expand on the idea of a python g-code generator.
[15:38:59] <rayh> You can. That is great Dan.
[15:39:13] <dan_falck> it's been years...
[15:39:48] <rayh> We approved a motion to point it at SWP's account as soon as the major content has been moved to joomla.
[15:40:00] <dan_falck> yes, the python G-code generator looks interesting
[15:40:09] <dan_falck> ok
[15:40:36] <dan_falck> I will call you in a bit with some info
[15:40:37] <rayh> Everything that's happening lately is really exciting.
[15:40:46] <rayh> Thanks Dan.
[15:41:23] <cradek> hi all
[15:41:44] <dan_falck> hello
[15:44:20] <rayh> Hi Chris
[15:53:23] <jepler> rayh: In general, I prefer creating g-code with interpreted languages like Python.
[15:54:03] <dan_falck> jepler: are you converting cp1 from tcl to python?
[15:54:21] <jepler> rayh: It's easy to get the basics of Python, but it's also a capable enough language to do things like output a spline curve with "sensibly sized" segments
[15:55:20] <jepler> dan_falck: I've never used cp1. I don't have in mind anything so literal as converting an existing program into Python.
[15:56:07] <dan_falck> It's got some handy routines in it.
[15:59:49] <rayh> I like the "conversational" sort of approach also.
[16:01:11] <rayh> except we can't call it conversational or we'd have to pay Hurco.
[16:01:34] <rayh> I like the idea of "geometric" programming.
[16:01:38] <SWP_Away> "interactive"
[16:01:42] <SWP_Away> SWP_Away is now known as SWPadnos
[16:01:49] <rayh> sellect a geometry
[16:02:04] <jepler> I'd rather write a program that creates output than use a program in the style of CP1.
[16:02:05] <rayh> oops typoing in progress.
[16:02:14] <SWPadnos> on the fly programming
[16:02:23] <jepler> but I may be out of step with what others want
[16:02:34] <SWPadnos> one of the big
[16:02:37] <SWPadnos> gah
[16:02:51] <rayh> How do you tell the python what sort of thing to write?
[16:03:19] <SWPadnos> one of the big features of the other hobby-level machine controllers is that they can directly import other files like DXF, STL, and BMP
[16:03:44] <SWPadnos> lots of people don't like to have to run many programs to get things done
[16:04:05] <jepler> rayh: Instead of a CP1 "MOP" you would have a Python function
[16:04:25] <rayh> No arg there.
[16:04:51] <jepler> the side-effect of the function would be to print the g-codes, usually to the destination file
[16:05:07] <rayh> You ought to ask Matt about cp2 sometime.
[16:05:52] <rayh> The MOP notion was a holdover from a NIST experiment with embedding xml in g-code comments.
[17:04:44] <rayh> Hi tom
[17:16:12] <tomp> rah: hi, sorry was following buzz on 5 axis other room
[17:45:51] <tomp> rayh: no word from teh 2 guys about the edm p/s, so building up my own fet p/s
[17:46:25] <tomp> rayh: board done, ballast res done , all mounted, building guts dc supply now
[17:47:03] <tomp> rayh: and the osc is being a problem, works now but isnt parallel input for emc yet
[17:48:46] <tomp> rayh: i gotta turn this off, too distracting & too little time left, bye
[17:53:46] <rayh> rayh is now known as rayh-away
[18:31:51] <alex_joni> rayh-away: when you get back, let me know
[18:34:47] <jmkasunich> I have a question about how to handle the PERIODs in the ini file
[18:35:11] <jmkasunich> today they are in seconds, the run script translates to nS and then insmods the motmod passing the periods as cmd line args
[18:35:30] <jmkasunich> I want to have the hal file loadrt motmod, and I need to pass it the periods
[18:35:36] <jmkasunich> I can do one of two thingsL
[18:35:45] <jmkasunich> 1) change the PERIOD in the ini file to nS
[18:35:53] <alex_joni> I see no problem with that
[18:36:00] <alex_joni> it's easier to convert
[18:36:12] <alex_joni> I usually get confused by 0.0000060
[18:36:33] <jmkasunich> 2) assume that the run script will still be the one invoking the ini and hal files, and let the runscript do the translation
[18:36:46] <jmkasunich> the runscript sets env vars BASE_PERIOD_NSEC
[18:36:52] <jmkasunich> which halcmd can read
[18:37:02] <alex_joni> I don't like #2
[18:37:27] <alex_joni> and there are little reasons for using #2 over #1
[18:37:36] <alex_joni> only keeping ini's compliant..
[18:38:27] <jmkasunich> ok
[18:41:26] <jmkasunich> hmm, 1000000 nanos isn't much better than 0.001 seconds (for servo)
[18:41:47] <jmkasunich> enough zeros that you have to count them to make sure you have it right
[18:41:59] <alex_joni> yeah, either way is hard
[18:42:20] <alex_joni> we need something like 1M
[18:42:22] <alex_joni> ;)
[18:42:29] <jmkasunich> yeah
[18:42:37] <alex_joni> or 10k3
[18:42:39] <jmkasunich> 1n, 1u, 1m, 1, 1K, 1M, 1G
[18:42:41] <alex_joni> ;-)
[18:43:04] <alex_joni> so you can say 530u ?
[18:43:08] <jmkasunich> right
[18:43:13] <alex_joni> or 2500n
[18:43:16] <jmkasunich> or 5.9u
[18:43:24] <alex_joni> yeah, but that makes it hard for hal
[18:43:39] <jmkasunich> yep
[18:44:43] <jmkasunich> forget it for now, maybe some other time
[18:44:49] <alex_joni> ok, 2.2
[18:45:26] <alex_joni> keep 10000 for now
[18:45:39] <jmkasunich> I suddenly have a bad feeling about changing SERVO_PERIOD in the ini file to nanos
[18:45:53] <alex_joni> how about all micros?
[18:46:05] <jmkasunich> are we absolutely sure its used only by motmod to set the thread periods?
[18:46:20] <alex_joni> but then you need to parse 0.xx
[18:47:41] <alex_joni> grepping for servo_period finds only motion.c and control.c
[18:47:58] <jmkasunich> uppercase is the way it is listed in the ini file
[18:48:18] <jmkasunich> and that grep find only the configs and emc (runscript)
[18:48:21] <jmkasunich> so maybe its OK
[18:48:22] <alex_joni> I grepped without case
[18:48:54] <jmkasunich> I was worried that perhaps some of the user space code also used it and expected seconds
[18:48:58] <alex_joni> yup, only ini's, emc.in and the 2
[18:49:09] <alex_joni> * alex_joni goes away for a while
[18:49:12] <jmkasunich> I wonder if COMM_WAIT is used anymore?
[18:49:14] <alex_joni> water is getting cold ;)
[18:49:27] <jmkasunich> that is in the motion section of the ini, and still in seconds
[18:49:35] <alex_joni> yes, by usrmotintf.cc
[18:49:49] <alex_joni> but it's also in emcmotcfg.h as a default value
[18:49:54] <alex_joni> and I don't think anyone changes it
[18:49:55] <jmkasunich> so that one can't be changed
[18:50:04] <alex_joni> I would vote for removing it from the ini
[18:50:07] <jmkasunich> I hate having a mix of seconds and ns
[18:50:52] <jmkasunich> I'm not going to remove it from the ini
[18:51:03] <jmkasunich> not unless we decide to remove the lines that look for it in the ini
[18:51:09] <jmkasunich> and I don't want to go there today
[18:51:14] <alex_joni> right
[19:21:11] <rayh-away> rayh-away is now known as rayh
[19:21:48] <rayh> alex_joni, I'm back!
[19:22:33] <jmkasunich> I think alex is away now
[19:23:03] <rayh> I see that.
[19:23:19] <rayh> We're still aiming for a branch tonight?
[19:23:29] <jmkasunich> if not tonight, then soon
[19:23:45] <rayh> Okay. I should have a working version of emccalib soon.
[19:28:41] <jmkasunich> I broke something
[19:28:52] <jmkasunich> base_period=[EMCMOT]BASE_PERIOD
[19:28:54] <jmkasunich> oops
[19:28:59] <jmkasunich> Error in startup script: can't read "activeAxis": no such variable
[19:28:59] <jmkasunich> while executing
[19:28:59] <jmkasunich> "set oldaxis $activeAxis"
[19:28:59] <jmkasunich> (procedure "axisSelectx" line 21)
[19:28:59] <jmkasunich> invoked from within
[19:28:59] <jmkasunich> "axisSelectx [ lindex $axiscoordmap 0 ]"
[19:29:01] <jmkasunich> (file "/home/john/emcdev/emc2head/tcl/mini.tcl" line 3540)
[19:29:21] <jmkasunich> I'm loading motmod using HAL now, and I get the above from mini
[19:29:53] <jmkasunich> I should have verified that it worked before I started changing stuff, think I'm gonna do that now
[19:32:08] <jepler> I saw that recently too
[19:32:27] <jepler> but then I ran the right .ini, and AXIS came up just fine
[19:32:32] <jmkasunich> yeah, I'm getting the same problem even when I let the run script load motmod
[19:33:15] <jmkasunich> ray: can you try a config that uses mini and see if it works?
[19:34:12] <rayh> Sure. Using today.
[19:34:34] <jmkasunich> does the sim config work?
[19:37:00] <rayh> mini fails here on a setup made from friday.
[19:37:14] <jmkasunich> same message that I posted a few minutes ago?
[19:37:17] <jepler> I did an update -D20060324 and it fails
[19:37:21] <jepler> same error
[19:37:32] <rayh> It worked properly on the previous system.
[19:37:43] <jepler> trying 0322 next
[19:37:45] <rayh> will update to today.
[19:38:48] <jepler> 0322 failed too
[19:40:14] <jmkasunich> mini.tcl got some pretty heavy edits on 3/12
[19:40:21] <jmkasunich> no changes since then
[19:41:09] <rayh> I believe that activeAxis is a local variable in mini.
[19:41:16] <jmkasunich> really heavy changes: Changes since 1.9: +1004 -484 lines
[19:42:38] <alex_joni> the activeAxis & similar were fixes I commited a while ago
[19:42:45] <alex_joni> when I fixed the multiple axes jogs
[19:43:01] <alex_joni> rayh: maybe you commited the mini from bdi-4.38 by mistake?
[19:43:43] <alex_joni> or maybe willingly?
[19:44:19] <jepler> 3/14 runs for me
[19:44:23] <jepler> (cvs update -D20060314)
[19:44:32] <jepler> so it doesn't seem to have been the changes on 3/12
[19:44:33] <rayh> I did commit an emc2 ized version of mini but it was working for a while afterword here.
[19:45:08] <jmkasunich> well if 3/14 works, then the problem isn't with mini directly, but something that it talks to
[19:45:12] <rayh> I can try to look at it after a bit.
[19:45:27] <rayh> That's where the local variable should exclude that.
[19:45:50] <jepler> 3/18 runs for me
[19:47:01] <jmkasunich> I'm building 3/23
[19:47:19] <jepler> 3/20 runs for me
[19:47:23] <rayh> you guys are much to fast for me.
[19:47:35] <rayh> Thanks for checking this out.
[19:47:36] <jmkasunich> checkouts over modem...
[19:47:44] <jepler> 3/21 doesn't
[19:48:13] <alex_joni> jepler: can you diff and point out some files?
[19:48:24] <jepler> alex_joni: I'm doing that now
[19:48:28] <alex_joni> jepler: thanks ;)
[19:48:32] <jmkasunich> looking at commit list archives
[19:48:33] <jepler> I don't have to rebuild so it's probably something in tcl/
[19:48:57] <alex_joni> rayh: does mini link/source emccalib ?
[19:49:16] <rayh> I don't think so. Let me look.
[19:49:46] <jmkasunich> if it did there would be a cvs entry for the change to mini.tcl
[19:49:52] <rayh> Yes it does.
[19:49:56] <alex_joni> jmkasunich: it already did
[19:50:01] <jmkasunich> oh
[19:50:35] <rayh> Let me comment that here. Line 874
[19:50:51] <jepler> I think it's the change in emccalib from 1.5 to 1.6.
[19:51:19] <jepler> at the same time, tkemc got rid of the source of emccalib, but mini didn't.
[19:51:19] <jmkasunich> question: why is mini invoking emccalib when emccalib isn't done yet?
[19:51:27] <rayh> Yep that's it.
[19:51:39] <rayh> I'll comment out the source and commit.
[19:51:43] <jmkasunich> ok
[19:51:45] <rayh> Thanks a bit guys.
[19:51:49] <jepler> just a bit?
[19:51:50] <jmkasunich> thanks everybody
[19:51:51] <rayh> thanks a bunch
[19:51:59] <jepler> oh ok *laughs*
[19:52:00] <jmkasunich> (I thought I did it with my motmod stuff)
[19:53:25] <alex_joni> jmkasunich: :-P
[19:53:31] <alex_joni> good thing you spotted it
[19:53:38] <rayh> wow was that an obsolete reference or what.
[19:54:05] <rayh> I put that in while working at Joe Martin's dining table before the sherline release.
[19:55:12] <alex_joni> rayh: I think it's these 3 lines that were removed
[19:55:13] <alex_joni> if {! [info exists activeAxis]} {
[19:55:18] <alex_joni> set activeAxis 0
[19:55:20] <alex_joni> }
[19:58:46] <jmkasunich> I think alex is right
[19:58:50] <jmkasunich> (and ray was right too)
[19:59:03] <jmkasunich> sourcing emccalib was a bug, and is better now
[19:59:09] <jmkasunich> but I still get the error
[19:59:49] <alex_joni> the same one?
[19:59:56] <jmkasunich> yes
[20:00:45] <jmkasunich> before I got a really screwy opening screen (with the narrow chips pic that appears on pickconfig, and some oddly placed buttons)
[20:01:01] <jmkasunich> removing the reference to emccalib fixed that, I got a normal mini screen
[20:01:17] <jmkasunich> but after a few seconds, it fails with the same error
[20:01:34] <alex_joni> * alex_joni tries it too
[20:01:37] <jmkasunich> Error in startup script: can't read "activeAxis": no such variable
[20:02:09] <rayh> Okay.
[20:04:57] <rayh> set activeAxis 0
[20:04:57] <rayh> is in there.
[20:05:53] <jmkasunich> as in "so it must be something else"? or as in "I just put it back in"?
[20:06:16] <alex_joni> jmkasunich: I read the message more carefully
[20:06:25] <jmkasunich> ok
[20:06:32] <rayh> I've got it.
[20:06:32] <jmkasunich> must be something else then ;-/
[20:06:33] <alex_joni> and [ lindex $axiscoordmap 0 ] is related to the ini
[20:06:52] <alex_joni> it contains a list with the axes found in traj
[20:09:38] <jmkasunich> I guess I should ask if anyone else had the problem now?
[20:09:57] <alex_joni> jmkasunich: I commited a short term fix to make it work again
[20:10:09] <jmkasunich> I removed a 5 second delay after installing the motmod, maybe that is the problem at my end (if nobody else has it)
[20:10:10] <SWPadnos_> SWPadnos_ is now known as SWPadnos
[20:10:16] <rayh> I'm a bit lost.
[20:10:26] <alex_joni> jmkasunich: no, not really
[20:10:27] <rayh> cause it's running here now.
[20:10:35] <alex_joni> it's not running here
[20:11:43] <alex_joni> well, it's running now, as I commented out that axisSelectx [lindex $axiscoordmap 0 ]
[20:12:05] <rayh> wait a minute. What the hell we doing.
[20:12:17] <jmkasunich> I dunno
[20:12:18] <rayh> And why
[20:12:32] <rayh> Did the traj section of the ini get screwed.
[20:12:32] <jmkasunich> heres what happened:
[20:12:37] <rayh> I am using my own ini
[20:12:59] <jmkasunich> 1) I changed the sim ini to use ns instead of secs for the periods (didn't touch the traj section)
[20:13:12] <jmkasunich> I changed the run script to _not_ insmod motmod
[20:13:22] <jmkasunich> I changed the core_sim.hal to loadrt motmod
[20:13:30] <jmkasunich> when I went to test it mini was busted
[20:13:44] <jmkasunich> but I have no idea whether my changes busted it or whether it was already busted
[20:13:49] <rayh> And emc_ini used the insmod
[20:13:54] <alex_joni> yeah, but you didn't commit that stuff
[20:14:02] <jmkasunich> no I didn't commit anything
[20:14:04] <alex_joni> so why was it busted here?
[20:14:27] <jmkasunich> well, that bad source had it busted for everybody
[20:14:29] <jmkasunich> that is now fixed
[20:14:36] <jmkasunich> but mine is still busted
[20:14:49] <jmkasunich> so is that because of my changes, or because of something else?
[20:15:03] <rayh> what " axisSelectx [lindex $axiscoordmap 0 ]" does is convert between axis names and axis numbers.
[20:15:18] <rayh> I really would hate to loose that ability.
[20:15:32] <jmkasunich> commenting that out was at best a temp fix
[20:15:38] <jmkasunich> I'd rather find the real problem
[20:16:05] <rayh> What config are you running mini with?
[20:16:10] <alex_joni> sim
[20:16:16] <rayh> let me try.
[20:16:23] <alex_joni> jmkasunich: I know, only meant a temp fix
[20:16:30] <alex_joni> to point out the problem
[20:17:47] <rayh> It's running here now with my two changes.
[20:17:56] <jmkasunich> two?
[20:18:16] <rayh> I just committed a second change.
[20:18:22] <jmkasunich> removing the reference to emccalib and ?
[20:18:34] <jmkasunich> oh, ok
[20:18:38] <rayh> moved the set active
[20:18:44] <rayh> phone
[20:18:49] <jmkasunich> CIA didn't do its thing yet
[20:18:50] <alex_joni> * alex_joni didn't see the second one
[20:20:12] <jmkasunich> CIA must be sleeping, just did a cvs up and theres a change
[20:20:16] <jmkasunich> gonna test now
[20:21:53] <alex_joni> I didn't see the change
[20:22:11] <jmkasunich> looks like its working now
[20:22:27] <alex_joni> make sure it wasn't my change you've seen
[20:22:36] <jmkasunich> duh, I bet it was
[20:23:57] <jmkasunich> alex: can you undo your change and commit
[20:24:02] <jmkasunich> thats just confusing the issue
[20:27:13] <alex_joni> ok, I can
[20:29:34] <rayh> Looks like the last change I made must have conflicted with alex cause I don't see mine at all.
[20:29:59] <alex_joni> it didn't get commited
[20:30:03] <rayh> Do we get any special recognition for confusing sf's cvs.
[20:30:10] <alex_joni> bet you got a message about not having the latest version
[20:30:46] <jmkasunich> ray: do a cvs up and look for M or C associated with mini.tcl
[20:30:48] <rayh> I didn't see it and the terminal has moved on but you are probably right.
[20:31:06] <jmkasunich> M means your change is still on your box but uncommitted
[20:31:09] <rayh> okay.
[20:31:15] <jmkasunich> C means yours and alex's collided
[20:31:50] <alex_joni> where is axisSelectx ?
[20:31:55] <rayh> I hate doing that without a make clean first but...
[20:32:16] <jmkasunich> ok, I just did a fresh checkout (which means none of my mods to the sim config) and it still fails
[20:32:36] <alex_joni> jmkasunich: that was expected
[20:35:00] <alex_joni> rayh: ok, I think I figured it out (activeAxis only gets set after the first call to axisSelectx which uses it)
[20:35:07] <jmkasunich> well it makes me feel a little better that the motmod stuff didn't cause some obscure bug
[20:35:09] <alex_joni> but why did this work before?
[20:35:24] <rayh> Merging differences between 1.11 and 1.13 into mini.tcl
[20:35:24] <rayh> rcsmerge: warning: conflicts during merge
[20:35:47] <jmkasunich> yeah, something happened during your prior commit
[20:35:50] <alex_joni> rayh: open the file and look for <<<< and >>>> messages
[20:35:52] <jmkasunich> search for <<<<<<
[20:36:06] <rayh> I moved the set activeAxis 0 to the top of the file.
[20:36:28] <jmkasunich> the stuff between/around <<< and >>> is the two versions that CVS couldn't figure out how to merge
[20:36:42] <rayh> not there at all.
[20:36:43] <alex_joni> * alex_joni gets it..
[20:36:43] <rayh> I
[20:36:53] <rayh> I've got a clean file that runs here.
[20:37:01] <alex_joni> so the set activeAxis 0 was actually beeing set in emccalib.tcl before
[20:37:11] <alex_joni> but now that ray started working on that it got busted
[20:37:30] <alex_joni> rayh: I can move the set activeAxis 0 on top of the file
[20:37:51] <alex_joni> although I suspect the axisSelectx which I first commented out doesn't do anything
[20:37:56] <alex_joni> as there is another call lateron
[20:38:08] <rayh> okay found it.
[20:38:41] <rayh> No it was set near the end before the loop that updates stuff.
[20:39:06] <rayh> I'll kill off my version, get the latest from alex and make the small change needed.
[20:39:24] <alex_joni> what? activeAxis? I know, but it was also set before that in emccalib.tcl, which is the reason why it worked until 03.21
[20:41:23] <rayh> Okay. In the new stuff from 4.38, there are a few menu items that wanted a value for activeAxis.
[20:41:59] <rayh> With the old mini, the references to activeAxis were in processes that hid them until run.
[20:42:52] <alex_joni> is it any better? the modded mini I mean
[20:42:58] <rayh> When emccalib was sourced, it tended to look at the activeAxis that was already working in tkemc.
[20:42:58] <rayh> I think.
[20:43:05] <alex_joni> it seems to startup slower..but I can be wrong
[20:43:20] <rayh> Yes there are quite a few improvements to the editing section.
[20:43:31] <alex_joni> when emccalib was sourced it looked for activeAxis, if it's not there it would have set it to 0
[20:43:38] <rayh> Paul found an editor that he could embed in there in place of the one I wrote.
[20:43:47] <rayh> Right.
[20:44:07] <rayh> The old version of emccalib was a bastard child that FredP cooked up.
[20:44:20] <rayh> supposed to be able to source or run stand alone.
[20:44:22] <alex_joni> I like that description ;)
[20:44:42] <rayh> Those things never worked very well so I wrote a stand alone that could be executed from tkemc.
[20:44:54] <rayh> or any where else.
[20:45:09] <rayh> I think cradek added a reference to it in axis.
[20:45:37] <rayh> This sourcing thing is probably why the file was not marked executable also.
[20:46:32] <alex_joni> well, then it should have been in scripts/ ;)
[20:46:39] <alex_joni> anyways, SF staff is real slow
[20:46:47] <alex_joni> they still haven't changed the permissions
[20:46:48] <alex_joni> :(
[20:46:51] <rayh> Ok it's broke now.
[20:48:17] <rayh> btw why do we get
[20:48:20] <rayh> Starting emc...
[20:48:20] <rayh> Version: 1.2
[20:48:20] <rayh> Machine: EMC-HAL-SIM
[20:48:20] <rayh> Version: 1.2
[20:48:21] <rayh> Machine: EMC-HAL-SIM
[20:48:23] <rayh> e
[20:48:39] <rayh> And now it works.
[20:50:22] <jmkasunich> I think the doubled messages are task and iocontrol both printing their version info
[20:50:24] <alex_joni> those 2 prints are one by iocontrol, and one by .. ?
[20:50:28] <alex_joni> task yeah
[20:50:36] <alex_joni> maybe they should spell their name too
[20:50:42] <rayh> ah okay.
[20:50:56] <rayh> Seemed like it was a relatively new thing.
[20:51:17] <rayh> Hey alex where did you comment out
[20:51:40] <jmkasunich> I'll fix those prints in task and iocontrol
[20:51:49] <jmkasunich> you guys stay on mini ;-)
[20:53:31] <alex_joni> rayh: I put it back (jmk said that) ;)
[20:53:49] <rayh> I didn't find it at all.
[20:54:10] <jmkasunich> he may have removed it before you got the version that was commented out
[20:54:33] <jmkasunich> I asked him to because it doesn't really fix the problem, just masks it
[20:54:46] <rayh> $Revision: 1.13 is what I'm looking at/
[20:55:49] <alex_joni> jmkasunich: it did fix it, because the axisSelectx call I commented out wasn't really used, there was another one lateron, after activeAxis was defined
[20:56:02] <jmkasunich> oh
[20:56:17] <jmkasunich> anyway, 1.12 is where he commented out the line, 1.13 is where he restored it
[20:56:46] <rayh> Okay. I'll move the one line and commit.
[20:59:33] <rayh> Says that it's there.
[21:00:01] <rayh> Thanks Alex for finding the dead lines. i really should work over the whole thing.
[21:01:48] <jmkasunich> ok, it works here now
[21:01:51] <jmkasunich> thanks guys
[21:01:55] <jmkasunich> cd ..
[21:02:17] <anonimasu> %
[21:06:40] <alex_joni> #
[21:18:39] <alex_joni> * alex_joni goes to bed
[21:18:41] <alex_joni> night all
[21:18:56] <alex_joni> it's 00:21 here (new time since last night)
[21:21:18] <rayh> Night alex. Thanks again.
[21:21:28] <alex_joni> no sweat.. night all
[21:21:28] <jmkasunich> goodnight alex
[21:21:41] <alex_joni> jmkasunich: can you let me know what got decided regarding release?
[21:21:49] <alex_joni> or maybe I'll ask chris tomorrow
[21:22:05] <jmkasunich> I'll be travelling tomorrow (work) so chris would be better
[21:22:18] <jmkasunich> I'm gonna try to get all the motmod stuff done tonight
[21:22:25] <jmkasunich> sim is working ;-)
[21:24:26] <alex_joni> ok, nice
[21:24:39] <alex_joni> then I can sleep safely
[22:04:29] <cradek> hi all, I'm back
[22:05:58] <jmkasunich> just in time ;-)
[22:06:15] <jmkasunich> I believe I've made the motmod changes to all configs
[22:06:27] <jmkasunich> sim and stepper work (aka mini and tkemc)
[22:06:34] <jmkasunich> tried max and it doesn't
[22:06:46] <jmkasunich> Traceback (most recent call last):
[22:06:47] <jmkasunich> File "/home/john/emcdev/emc2head/bin/axis", line 61, in ?
[22:06:47] <jmkasunich> import gcode
[22:06:47] <jmkasunich> ImportError: /home/john/emcdev/emc2head/lib/librs274.so: undefined symbol: _Z22START_SPEED_FEED_SYNCHd
[22:07:53] <jmkasunich> does that line mean anything to you?
[22:07:54] <cradek> you'll have to update axis
[22:07:57] <jmkasunich> ah
[22:08:20] <cradek> it has to be updated every time the canon api changes
[22:09:08] <jmkasunich> cvs up in my axis checkout, followed by make in the emc tree? or do I need a configure in the emc tree first?
[22:09:33] <jmkasunich> dang
[22:09:37] <cradek> if you put the axis directory in emc2head/src, it will build automatically
[22:10:03] <jmkasunich> the axis CVS checkout tree is outside emc2head, but theres a symlink to it in emc2head/src
[22:10:19] <jmkasunich> I didn't want to next cvs checkouts, seemed scary
[22:10:21] <cradek> that should be fine
[22:10:26] <jmkasunich> s/next/nest
[22:10:34] <jmkasunich> I got a conflict on my axis cvs up
[22:10:53] <cradek> odd, have you been changing axis?
[22:10:59] <jmkasunich> in setup/monkeypatch.py
[22:11:21] <jmkasunich> I think jepler had me change a line in there to prevent a rebuild of axis every time I built emc
[22:11:34] <jmkasunich> don't remember the details, checking the conflict now
[22:12:03] <jmkasunich> yep, he did the same thing in the official version
[22:12:10] <jmkasunich> thus the conflict
[22:14:58] <jmkasunich> all better now
[22:16:14] <jmkasunich> heh, emc starts up 5 seconds faster now
[22:17:08] <jmkasunich> there was a sleep 5 after the insmod of motmod (dunno why, sleep is needed sometimes after a user space module is loaded before you try to linksp things to it, but loadrt/insmod doesn't return until its done)
[22:17:42] <cradek> I think I did that
[22:17:55] <cradek> I would often get "that hal pin doesn't exist" type errors without it
[22:18:05] <jmkasunich> from motmod?
[22:18:18] <jmkasunich> iocontrol pins yes
[22:18:29] <cradek> yeah maybe it was iocontrol
[22:18:31] <jmkasunich> but motmod pins should all exist when the insmod returns
[22:18:31] <cradek> it's been a while ago
[22:18:43] <jmkasunich> theres a sleep 1 after iocontrol is loaded
[22:21:44] <jmkasunich> 26 files in 16 dirs... mods to the sample configs are _such_ fun....
[22:22:40] <rayh> We should make the parport configs read the location from ini as well.
[22:22:59] <jmkasunich> _now_ he tells me ;-)
[22:23:33] <rayh> sorry. your commit reminded me of something I needed to do.
[22:23:39] <jmkasunich> lol
[22:24:01] <jmkasunich> * jmkasunich goes back to messing with LVM
[22:27:41] <jmk_cf> uh oh, clones...
[22:29:42] <jmk_cf> actually, now seems like a good time for dinner
[22:39:29] <steves_logging> steves_logging is now known as steve_stallings
[22:55:46] <rayh> SWPadnos, You around.
[23:26:05] <SWPadnos> rayh, I'm back now
[23:53:49] <rayh> Hi SWPadnos
[23:53:59] <SWPadnos> hi there
[23:55:06] <rayh> We need to move the DNS for linuxcnc in the next week. If you get me the proper whatever to your location I can arrange that.
[23:55:49] <SWPadnos> all they have to do is change the DNS over to ns1.dreamhost.com (with ns2 and ns3 as backups)
[23:56:13] <SWPadnos> I'll need the information on wiki.linuxcnc.org, so they can point it at sourceforge correctly
[23:56:41] <rayh> Ah I have no clue about the wiki.
[23:56:57] <rayh> Think steve_stallings might?
[23:57:05] <SWPadnos> I may be able to get the info from a DNS lookup, or from Steve
[23:57:09] <SWPadnos> yes, he has it
[23:58:27] <rayh> So there's nothing beyond ns1.dreamhost.com no reference to cncgear or any such.
[23:58:54] <steve_stallings> you need to add an alais or CNAME record that equates "wiki" with "vhost.sourceforge.net
[23:59:00] <SWPadnos> ok
[23:59:19] <SWPadnos> I can do that now. you can test it by putting ns1.dreamhost.com at the top of your resolv.conf file
[23:59:33] <SWPadnos> (except on Windows ;) )