#emc | Logs for 2004-12-02

[00:20:23] <robin_z> meep?
[00:20:45] <paul_c> Not a chance this time of night. ;)
[00:20:55] <robin_z> damn ...
[00:21:48] <paul_c> back in a bit.
[00:23:24] <robinz> bloody firewall
[00:24:02] <robinz> todays task is trying to figure out how to build a powder coat booth for $ trivial
[00:24:19] <robinz> probably using an old office fan and some washing up liquid bottles
[00:26:03] <paul_c> a pair of Val's old knickers, a toilet roll, and some sticky backed plastic.
[00:26:28] <robinz> thats the sort of thing ...
[00:26:37] <robinz> but without the smell of fish, if possible
[00:29:01] <robinz> there went one of me ...
[00:30:19] <robinz> I'm trying to find some suitable filters, cheap
[00:30:30] <robinz> pleated paper I guess fo rthe final filter
[00:30:51] <robinz> I was wondering about just stacking a few dozen of the cheapest car air filters I can find :)
[00:30:52] <paul_c> Truck air filters
[00:30:56] <robinz> cheap?
[00:31:52] <robinz> I figured they;d be in the several undred pound category, no?
[00:31:53] <paul_c> Or what about a sheet of the foam they use for 80mm fans
[00:32:20] <robinz> foam I fgured would be too leaky
[00:32:37] <robinz> I thought pleated paper, or maybe fine weave cloth
[00:33:50] <robinz> I have a desire NOT to fill the factory with powder coat!!
[00:35:04] <robinz> you think truck air filters might be cheap though??
[00:41:34] <paul_c> Might be worth a look
[00:41:57] <paul_c> It really depends on the volume of air you need to filter...
[00:42:26] <paul_c> An electrostatic filter might be better.
[00:45:50] <paul_c> Hi Ray
[00:45:57] <rayh> Hi Paul
[00:46:12] <paul_c> Burn No. 5 on the go.
[00:46:30] <rayh> Getting closer?
[00:47:03] <rayh> I've got 208 fragments in lost and found on the suse box. Think I need to start over?
[00:48:06] <paul_c> 'pends where the fragments came from, and more importantly, How.
[00:48:50] <rayh> There are only a dozen or so files in /etc now.
[00:49:19] <paul_c> That doesn't sound good
[00:57:20] <les> robin: still there?
[00:59:02] <les> guess not...pity...could tell him about GEMA powder booths
[01:02:39] <les> nothing special...pressurize a little room...pleated paper filters...keep a shop vac around
[10:44:59] <alex_joni> anybody around?
[11:26:27] <alex_joni> guess not
[11:58:28] <alex_joni> good morning paul !!!
[11:58:45] <paul_c> Yo alex_joni
[11:59:07] <alex_joni> I just had a great idea last night
[11:59:16] <alex_joni> you'll probably say it's a bunch of bull.. ;)
[11:59:22] <alex_joni> but I'll tell you anyhow
[11:59:45] <alex_joni> how about this: what if hal modules are smaller things
[11:59:56] <alex_joni> e.g. for STG not a driver for the whole board
[12:00:03] <alex_joni> but instead drivers for the components
[12:00:15] <alex_joni> hal_ls7166 , hal_82c55, etc
[12:00:25] <alex_joni> and those can be used for many hardware parts
[12:00:42] <alex_joni> just need a base_address for those, and maybe a few more things
[12:01:46] <paul_c> Base address, byte boundaries, IO or memory mapped....
[12:02:25] <alex_joni> too complicated ?
[12:02:45] <paul_c> Some things are best done in C
[12:02:51] <alex_joni> I thought maybe there's a way to have a stg.hal
[12:04:10] <alex_joni> well... on the other hand it's not possible to load a module more than once
[12:04:48] <alex_joni> so it's pretty complicated to do the param_parsing inside the module (CFG="200 201 202 203")
[12:05:29] <paul_c> Pass smple numbers NOT text that requires parsing in kernel space.
[12:06:15] <paul_c> manipulating strings in kernel space is all wrong
[12:08:50] <alex_joni> I know.. I don't like the cfg from hal_parport
[12:09:45] <paul_c> so why follow the example.
[12:14:06] <alex_joni> I don't want to...
[12:14:27] <alex_joni> I was only checking if it could be done with more components opposed to a single driver file
[12:14:35] <alex_joni> and what that would imply
[12:18:39] <paul_c> I 'spose taking it to it's logical conclusion, turn every function within EMC in to a HAL module...
[12:23:05] <alex_joni> well.. not that extreme
[12:23:06] <alex_joni> ;)
[12:23:24] <alex_joni> but I agree that keeping stuff in c has it's advantagies
[12:23:40] <alex_joni> if it's GPL it can be used in other drivers too...
[12:23:46] <alex_joni> right?
[12:24:27] <paul_c> Licensing is secondary to the efficient use of code.
[12:31:27] <alex_joni> I agree...
[12:46:15] <paul_c> Another thing to bear in mind...
[12:46:51] <paul_c> gcc can do a far better job of linking and optimising code than most of us.
[12:47:23] <alex_joni> how do you mean this?
[12:53:47] <paul_c> Write a series of functions that use small (2-3 line) function calls...
[12:54:07] <paul_c> bung in some large loops and switch statements
[12:56:02] <paul_c> gcc will inline many of the small functions, unroll loops, and skip any unreachable code
[12:56:21] <paul_c> Depending on the level of optimisation called for.
[12:59:25] <paul_c> Splitting the short functions off into a separate module "just because we can"
[13:00:01] <paul_c> and you loose any chance of opimisation.
[13:00:50] <paul_c> but, hey, a modern processor can do a jump in less than a nano-second...
[13:06:56] <paul_c> Arg... Don't have my Intel book to hand, so can't give a timing for a typical test&branch instruction..
[13:11:55] <jepler> paul_c: they barely publish info like that anymore, it seems like.
[13:13:08] <jepler> paul_c: but with many-stage pipelines, a mispredicted branch might take tens of cycles, or >5ns on a 2GHz machine.
[13:14:31] <paul_c> The size of the cache(s), pipelines, speed of main memory...
[13:31:32] <alex_joni> yeah.. it's all into non-RT direction nowadays
[13:55:11] <CIA-4> 03Zathras 07BDI build system * 10Babylon Cluster/comps.xml: File changed. New revision:picax.xml
[14:37:38] <alex_joni> paul: around?
[14:40:37] <alex_joni> guess not..
[14:40:39] <alex_joni> jepler?
[14:43:28] <jepler> hi, I'm here
[14:44:46] <alex_joni> * alex_joni is reading about AXIS
[14:45:00] <jepler> uh oh, somebody found our website
[14:45:05] <alex_joni> lol
[14:45:07] <alex_joni> yeah ;)
[14:45:15] <jepler> so what do you think?
[14:45:16] <alex_joni> but I'm not gonna say anything :D
[14:45:19] <jepler> oh good
[14:45:30] <alex_joni> I think .. WTF is it hidden?
[14:45:42] <alex_joni> it should be out there...
[14:45:42] <alex_joni> :D
[14:45:45] <alex_joni> it's GREAT
[14:55:07] <alex_joni> hello ray
[14:55:31] <rayh> Hi alex.
[14:55:53] <rayh> I abandoned the SuSE after two crashes.
[14:56:05] <alex_joni> have you been following the USB stuff?
[14:56:26] <alex_joni> the USB-parport adapter discussion on the users list
[14:56:28] <rayh> The second ate most of /etc and tossed 200+ fragments into lost+found.
[14:56:40] <rayh> At a glance only.
[14:57:04] <rayh> What are you thinking about USB?
[14:57:08] <alex_joni> huh... never had a single crash with SuSE
[14:57:20] <alex_joni> well.. I think USB is pretty useless for RT stuff
[14:57:34] <alex_joni> but then I found this: http://www.low-cost-cnc.com/ll_cncarchitecture_big_cnc.html
[14:57:48] <rayh> Yes. At least the way folk are thinking about it.
[14:58:10] <paul_c> rayh: Will check on the missing dep.
[14:58:48] <alex_joni> ray, paul: is there a need for a iso's-mirror?
[14:59:16] <paul_c> Not yet - Still at the beta stage.
[14:59:24] <alex_joni> I meant generally
[14:59:37] <alex_joni> also for the BDI2.x and BDI-TNG & Live
[15:00:01] <paul_c> TNG is unsupported and needs to be retired
[15:01:52] <alex_joni> ok.. then let me know when the need is there (I got the bandwidth & the disk space covered)
[15:02:38] <rayh> Sounds good to me. That way we have an (almost euro) mirror.
[15:03:13] <rayh> When does ro expect to join?
[15:03:22] <alex_joni> 2007+
[15:03:24] <alex_joni> I think
[15:03:35] <paul_c> Looking at slimming Live down so that it will run on a minimal system without many of the desktop apps
[15:04:01] <rayh> That's not so far off.
[15:04:11] <alex_joni> * alex_joni left ...
[15:08:35] <paul_c> Found the problem with the kernel not installing...
[15:09:48] <rayh> Good.
[15:11:13] <rayh> What was it missing?
[15:11:44] <paul_c> It was a case of what wasn't missing....
[15:12:05] <paul_c> I had a whitespace in the XML tag
[15:12:08] <rayh> Oh. That's much worse.
[15:17:09] <paul_c> are you sure that missing dep was libneon and not libneon24 ?
[15:38:25] <rayh> Might have been. It went by pretty fast.
[15:39:38] <paul_c> hehehe.... Looking at the cia stats....
[15:39:48] <paul_c> Zathras is No. 1
[15:40:27] <paul_c> alex_joni is 2nd
[17:33:51] <paul_c> * paul_c posts to emc-user commenting about the use of legacy units (the Inch).
[17:40:32] <jepler> I found Tom Hubin's remarks a little bit offensive, since I'm sure patches that fix metric behavior would be addressed, and someone committed to working on metric would be accepted as a developer.
[17:40:42] <jepler> (says the guy who's never committed anything to EMC yet)
[17:41:34] <jepler> er, "patches .. would be incorporated"
[17:44:19] <paul_c> Priorities - I do the BDI, other things are lower down the list.
[17:45:26] <jepler> cradek: since you just looked at emctaskmain.cc, will you fix the interp_state-when-starting-by-step bug?
[17:45:36] <cradek> jepler: urgh
[17:45:39] <jepler> paul_c: and we thank you for doing BDI!
[17:47:16] <cradek> so this is just a problem with the tcl backplot?
[17:47:50] <paul_c> I think so.
[17:49:16] <cradek> unsurprisingly, gplot and emcplot3d both get this right...
[17:49:40] <cradek> g0x0y0
[17:49:40] <cradek> g1x1
[17:49:40] <cradek> g21
[17:49:40] <cradek> g1y25.4
[17:49:40] <cradek> g20
[17:49:43] <cradek> g1x0
[17:49:45] <cradek> y0
[17:49:56] <cradek> at least they display a square when I do this
[17:50:19] <paul_c> try it with G20, then again with G21
[17:50:46] <cradek> you mean g20 before the first g1x1?
[17:51:53] <paul_c> yes
[17:52:35] <cradek> huh, mini draws a square too
[17:52:41] <cradek> what am I missing here?
[17:54:50] <cradek> I agree with his comments about G92
[17:55:13] <cradek> if you can't use fixed tool offsets, you're screwed
[17:55:30] <paul_c> the image should be scaled by 25.4...
[17:55:57] <cradek> I've never used the tcl backplot, so maybe I don't know how it works
[17:56:03] <paul_c> Agreed the G92 needs looking at...
[17:56:22] <paul_c> I have some files that claim to fix it, but
[17:56:59] <cradek> I bet you *could* do it using run-from-line
[17:57:04] <cradek> it sucks but you could do it
[17:59:05] <paul_c> I have some modified files, but it is a bloody nightmare trying to pick out what changes have been made to the actual code
[17:59:31] <cradek> lunch break
[17:59:37] <paul_c> if/when I can sort out the mess, I'll try to commit some of it.
[19:16:52] <rayh> cradek: If you are using a current mini gui, the g20 and g21 have been fixed in the plotter.
[19:17:06] <cradek> rayh: ok, that explains why it worked for me
[19:17:29] <rayh> I'm not certain how close the two squares will come to overlapping.
[19:17:45] <cradek> I changed to metric for one corner, so there is only one square
[19:17:52] <cradek> it looks fine
[19:18:18] <cradek> do you know if Tom has a sherline?
[19:18:31] <rayh> Okay. You could test it by making a 1 inch square followed by a 25.4 mm square.
[19:18:36] <rayh> I believe so.
[19:19:00] <rayh> I could rustle around and try to find my list of S customers.
[19:19:06] <les> what's this g92 stuff again?
[19:19:09] <cradek> nah, don't bother
[19:19:19] <cradek> les: http://www.linuxcnc.org/dropbox/g92test1.pdf
[19:19:32] <les> ray, you proved that it works proprly
[19:20:40] <les> I use it a lot
[19:21:12] <rayh> As long as you stay within the NIST parameters it's okay.
[19:21:27] <les> the only thing that was problematic was the gui sometimes not displaying an offset
[19:21:46] <les> so I just always start with a g92.1
[19:21:50] <les> no problems
[19:21:54] <rayh> Yes. I found that it was a bit more than just the gui.
[19:22:06] <les> oh I see
[19:22:20] <rayh> So did one of my customers who drilled a hole in his work vice.
[19:22:22] <les> message was not reported?
[19:22:31] <les> nml
[19:23:18] <rayh> Now I can't remember just where the issue was.
[19:23:39] <les> oh well...just took delivery of over $2000 worth of prime cherry
[19:23:57] <les> emc will turn 75% of it into sawdust.
[19:24:03] <rayh> I feel a bit like the old beer commercial announcer. "I've been drinking xx ale for as long as I can remember -- which is about three days.
[19:24:15] <les> haha
[19:25:54] <les> well late lunch break for me...later...and I have a bad case of CRS as well!
[19:33:12] <rayh> Looks like our bug and feature tracker has a bit of a issue with forwarding to the dev list.
[19:34:22] <paul_c> yup - Thought I had fixed it...
[19:37:02] <rayh> I'm trying to use supermount like live. Modprobe supermount. Anything else.
[19:38:07] <paul_c> /cdrom /cdrom supermount auto,users,ro,fs=auto,dev=/dev/hdc 0 0
[19:38:14] <paul_c> (for fstab)
[19:39:01] <paul_c> and add supermount to the lis in /etc/modules
[19:39:18] <rayh> I've got /media/cdrom.
[19:39:40] <paul_c> /cdrom /media/cdrom supermount auto,users,ro,fs=auto,dev=/dev/hdc 0 0
[19:39:57] <rayh> k
[19:40:17] <paul_c> Hmmm... that doesn't look right...
[19:40:41] <rayh> No /dev/cdrom
[19:41:36] <paul_c> Is the drive hdb or hdc ?
[19:41:49] <rayh> made a link scd0 ->cdrom
[19:42:05] <rayh> or should i use hdc?
[19:42:30] <rayh> burner it is.
[19:43:37] <paul_c> Is ide-cd being loaded along with scsi-ide ?
[19:44:21] <paul_c> cdrom /media/cdrom supermount auto,users,ro,fs=auto,dev=/dev/sr0 0 0
[19:44:45] <paul_c> or) cdrom /media/cdrom supermount auto,users,ro,fs=auto,dev=/dev/scd0 0 0
[19:46:13] <rayh> I don't see either ide-cd or scsi-ide but scd0 will work with a manual mount.
[19:55:03] <paul_c> must have compiled them into the kernel then..
[19:55:32] <rayh> I'm not getting either floppy or cdrom with any of the lines.
[19:56:16] <paul_c> supermount loaded ?
[19:56:39] <rayh> yep.
[19:57:14] <paul_c> 2.4.27 kernel ?
[19:58:11] <paul_c> do a "mount -a"
[19:59:48] <rayh> none already mounted or /dev/pts busy
[20:00:13] <rayh> according to mtab, devpts is already mounted on /dev/pts
[20:00:33] <rayh> fs type sysfs not supported by kernel
[20:01:34] <paul_c> get the same warnings here...
[20:02:06] <paul_c> mount should tell you if the CD is now accessable.
[20:02:58] <rayh> I could mount before supermount but with the mount command.
[20:05:14] <paul_c> does "mount /media/cdrom" throw any errors ?
[20:07:16] <rayh> permission denied
[20:07:32] <rayh> I'm root
[20:07:40] <paul_c> running it as root ?
[20:07:51] <rayh> yep
[20:08:24] <paul_c> same error..
[20:08:44] <rayh> Ah. A reboot with fstab in place and supermount in the /etc/modules got it going.
[20:08:59] <paul_c> a bit drastic
[20:09:35] <rayh> Yes it was.
[20:12:21] <rayh> I will not allow me to unmount manually now.
[20:13:49] <paul_c> Once supermount is running, it looks after mounting/unmounting
[20:14:27] <paul_c> although.... "umount /media/cdrom" should still work...
[20:14:31] <rayh> Okay. If I remove the media and look then install media it works properly.
[20:15:51] <rayh> The supermount works much better if there is a directory above the mount directory.
[20:16:18] <rayh> When you hit that directory with the browser it looks at the cd or floppy.
[20:16:29] <paul_c> Can you try something for me....
[20:16:52] <rayh> Go away;)
[20:17:02] <paul_c> pop the BDI-3 cd in, and run (as root)...
[20:17:14] <paul_c> apt-cdrom -d /media/cdrom
[20:17:36] <paul_c> apt-cdrom -d /media/cdrom add
[20:18:09] <rayh> need to mount it first?
[20:18:32] <paul_c> nope
[20:19:24] <rayh> Got the apt-cdrom help message. Want an add
[20:19:46] <paul_c> try the second incantation.
[20:20:02] <paul_c> * paul_c forgot the add
[20:20:38] <rayh> Okay it did something/asking if I want to repeat for other cds
[20:21:13] <paul_c> there isn't any more...
[20:21:18] <asdfqwega> Afternoon,guys
[20:21:43] <paul_c> Now try "apt-get install emc"
[20:21:52] <asdfqwega> * asdfqwega just bumped 2 20Gig HD's off the desk and onto the (concrete) floor
[20:22:04] <paul_c> It should ask for the CD
[20:22:22] <rayh> emc is already the newest version
[20:22:41] <learath> hope they didn't have anything important ont hem
[20:22:42] <rayh> Darn. need to get a net going.
[20:23:04] <paul_c> OK... now "apt-get install synaptic"
[20:24:07] <rayh> not available, referred to by another package.
[20:24:31] <rayh> package synaptic has no installation candidate.\
[20:24:53] <paul_c> Oh drat..
[20:26:41] <rayh> That cd access broke supermount. Had to mount /media/cdrom
[20:26:50] <paul_c> better make sure I get that right on the next CD...
[20:27:19] <paul_c> * paul_c wondered if apt-cdrom would unmount the CD
[20:27:42] <rayh> Yes it did.
[20:27:49] <rayh> Or said it did.
[20:29:17] <paul_c> Need to try playing around with autofs and see if that causes any problems.
[20:29:36] <rayh> Okay. How?
[20:30:22] <paul_c> can I write a micro-HOWTO a bit later ?
[20:30:26] <jepler> asdfqwega: ow
[20:30:34] <jepler> asdfqwega: hope you don't care too much about whatever data was on them
[20:30:54] <paul_c> (flog them on ebay)
[20:30:54] <cradek> jepler: the first junk power supply I picked up said 12V/10A
[20:31:02] <rayh> Certainly
[20:31:04] <jepler> paul_c: is it possible to "apt-get" on bdi live rc46 without doing a hard-drive install?
[20:31:08] <jepler> cradek: sold!
[20:31:13] <cradek> jepler: no kidding
[20:31:48] <paul_c> jepler: Yes, but it soaks up memory.
[20:31:48] <asdfqwega> Who's needing power supplies?
[20:31:51] <jepler> but my "cheapest CNC retrofit kit" just got a lot more expensive if I ship an AT power supply with it.
[20:32:00] <rayh> I see there is a CVS directory with your info on it in the cdrom.
[20:32:04] <jepler> paul_c: how? I get an error about free disk space on /var/apt/something
[20:32:34] <paul_c> You need loads more memory then..
[20:32:47] <jepler> paul_c: It's a 1-gig RAM machine
[20:32:51] <paul_c> rayh: I know about the CVS dir...
[20:33:39] <asdfqwega> Need memory: Just steal a few sticks out of your (gran)kids hot-rod gaming box
[20:34:50] <rayh> Can't you make a ramdisk on the hd and use that for rc46?
[20:34:59] <paul_c> jepler: I suspect translucency might be causing problems - It is something that has come up on the Morphix list
[20:36:16] <jepler> Yeah, I *think* that apt is just looking at "df" which says there are 0 blocks free on that partition
[20:37:14] <jepler> I mounted an ext3 partition on the hard drive and told yum to use that for temporary files, but then I got some error about cross-device hard links (my bug reporting stinks, I didn't write anything down)
[20:37:33] <rayh> ouch
[20:39:00] <paul_c> You could try overlaying the hd on top of / with translucency.
[20:39:27] <paul_c> (might get a bit messy in places...)
[20:40:03] <jepler> can you tell me a little more or point me at a document about this?
[20:40:19] <paul_c> gimme a sec...
[20:40:55] <paul_c> http://sourceforge.net/projects/translucency
[20:46:50] <paul_c> time to throw some food in the oven
[20:47:45] <jepler> thanks for the link
[22:12:20] <tbl> hio
[22:14:41] <paul_c> Cooo... A Mike
[22:24:05] <paul_c> so is tbl going to sit there all night lurking ??
[22:44:17] <tbl> sup paul
[22:45:57] <paul_c> running another test install...
[23:03:44] <robin_sz> are we having fun yet?
[23:04:05] <paul_c> Yahoooo....
[23:12:03] <jepler> I'm having a great time
[23:47:40] <asdfqwega> I'm having as much fun as a SM dentist