did memtest find bad memory?
here is the thing - I uninstalled the 512 and it worked.
i installed a differnt 512mb memory and it failed
swapped the location of the 512 and the 1gb same isssue (no start)
runs fine with 1gb
so it just plain doesn't like having 1.5G, regardless of which chips are in which socket
now - I installed ubunto on my workstation (dell pentium 4) with 1.256gb
it did the same thing - if I remove the 256mb it ran
this is a totally different computer?
yes other then they where both pentum 4 computer
you may have found a subtle but in linux somewhere
(or in the pr, for all I know)
cradek tried doing a make -j last night (compiles all files in parallel)
right - it runs just fine with just 1gb
that loads yp memory with lots of copies of gcc, and he started getting segfaults and other errors once he filled memory
so its not a emc, rtapi, or even rtai thing, its a "memory over 1G" thing
its just that rtai allocates shared memory at the top, so it shows up sooner
I can't remember what credak said - let me look at the history
anyway, what is important is that you are running
yes - thanks
I just got home from work, need to have dinner with my wife, then we're going out
but my curiosity was killing me, had to find out what happened ;-)
have a good night - the computer is on if you still want to look at it. with 1.5gb
I won't be online tonight
ok, I thought I was dreaming..
must be so ;)
SWPadnos: do you know anything about tape drives?
I just bought one, in fact
back in a minute
so - looking for SCSI tape drives?
local stores are pretty scarse
they're expensive, that's for sure
ok, that's pretty small
in fact, it's very small
found a SONY, DDS4 SDT 11000
for around 900$
that's kind of expensive (though it may be your location)
great, 350$ in the US
an AIT drive is higher capacity and roughly half the price
now that's a rip-off
ok - not half that price ;)
[00:58:04] <alex_joni> http://www.pacificdatastorage.com/sdt-11000.asp
I think around 500$ it's ok
yep - I've seen those
DDS4 is pretty old technology, and there isn't much of an upgrade path
how long do you expect to use this machine?
AIT tapes seem awfully expensive
pretty long , why?
the tapes are
I see DDS3 12G tapes for 15$
because you may have a hard time finding replacement hardware that will read DDS4 in 5 years
if something breaks
and AIT1 35GB for around 250$
that's probably IDE, at that price
that's almost 20 times more expensive
no, I mean the tape, not the drive
wait - for the cartridge?
that's a 10-pack, at least
olr a true rip-off
not at all, 1 pcs
[01:00:40] <alex_joni> http://www.flamingo.ro/produs.asp?id=4066
exchange rate is about 3:1
well, that's about 10x what it shuold cost
[01:02:46] <alex_joni> http://cgi.ebay.ch/Sony-SDT-11000-SCSI-tape-drive-40-GB-LIKE-NEW_W0QQitemZ8760106406QQcategoryZ8056QQcmdZViewItem
100 EUR, that's better
[01:03:15] <SWPadnos> http://www.newegg.com/Product/Product.asp?Item=N82E16840999208
that's closer to what I'm used to (cost-wise)
$50 for a 200G tape is much better ;)
that's too big for now
that was just for the price comparison
yeah, rip-off :(
but you should think about that as well, in terms of future storage growth
DDS goes up to DDS4, which is 20/40G, I think
oh my, this is nice: http://cgi.ebay.de/Quantum-ATL-P3000-Band-Roboter-Tape-Library_W0QQitemZ8763847869QQcategoryZ79407QQrdZ1QQcmdZViewItem
AIT starts at 35/90G, and goes up from there
Tape Drive: 12 x DLT-7000 (35/70G)
I would have 11 for spare ;)
although 600kg ;)
I have an AIT-2 drive I can sell you, but make sure you have HVD SCSI :(
I have U160
that can be either LVD/SE or HVD
though HVD may have staerted with U320
is there a way to tell?
the controller and devices should say what interfaces they support
don't plug LVD devices into an HVD controller (though the other way around seems to do no harm ;) )
[01:10:13] <SWPadnos> http://cgi.ebay.com/Qualstar-TLS-412600-AIT-3-600-Slot-Tape-Library_W0QQitemZ8766071084QQcategoryZ51090QQrdZ1QQcmdZViewItem
if you want a little extra capacity ;)
yeah, but only 2 drives in it
they probably have aggregate speed higher than the 12 DDS drives
that's something else to look at - how long a backup takes
AIT2 can write 6MB/second uncompressed, or 15 compressed
I think it's double for AIT3
if you're backing up 20G, AIT will take around an hour (+/-)
I think I can get this one pretty cheap : http://cgi.ebay.de/SONY-SDX-500C-SCSI-LVD-SE-68Pin-AIT-2_W0QQitemZ8767735977QQcategoryZ8056QQrdZ1QQcmdZViewItem
I think DDS is in the 1-2M/sec range (if that), so a 20G job will be in the 3 hour range
that's what I have
but he thinks it's broken, that's why he's selling it ;)
probably it will go for under 20EUR
[01:14:19] <SWPadnos> http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=8741312847
that's the exact one I bought
still one left
media: AIT 170M, AIT2 230M ?
that drive isn't quite as good as the SDX-510C, which gets 2.3:1 compression (instead of 2:1 for this drive)
it will read/write any AIT media with a lower number than the drive, so AIT1 or AIT2
and if you get a bigger HD, you can buy an AIT3 or AIT4 drive, and you'll still be able to read/write these tapes
so - the things to look at when buying a tape drive:
1) interface (IDE, SCSI LVD, SCSI HVD ...)
3) write speed
4) media costs
5) upgrade path
is there a way to tell what kind of SCSI I have?
6) media compatibility with future drives
do you have the card?
it's an onboard SCSI
on a dual Xeon
ok. U16o you said?
hang on looking for lspci
I'm not sure the chip type will tell you, though it may
Adaptec AIC-7899P U160/m
same as I have on my AHA39160
which happens to be LVD/SE, but I'm not sure that means yours is
can you tell from the HDD type?
the HDD should be labeled
or the model number may tell you
looks like it's LVD
LVD / Ultra-2 SCSI / Single Port
and only 80 MB/sec (Ultra2)
ok, so that tape drive I linked to should work
50 EUR for a 100/260G AIT-3 tape on ebay
that's a lot (and won't work in that drive)
you need the SDX2-50C
though the SDX2-36C will also work, but has lower capacity
I'll look around ebay.de see what I can find
ok. expect to pay a fair amount for the media, by the way
a normal backup plan uses roughly 10-12 cartridges, and you save one each month (or quarter)
but that's a business-y way to do things. for home use, you can do a bit less ;)
surprised to see you this late
seen you & cradek took off where I left skunkworks
that was a weird one
jmkasunich: I had some hard days lately, and slept this afternoon
still is ;)
yeah, but not a rtapi weird one
I wonder if there is some kernel config that affects how it handles memory over 1G?
there are several options
the normal split is 1G kernel / 3G user
that is the virtual address space
I'm talking about physical ram
it uses physical RAM over 1G without issue, most of the time
skunkworks apparently has TWO different boxes, both of which run emc2 fine when they have 1G, but fail when more is installed
same installation type (BDI / ubuntu / ...)?
the symptom is writes to an address like 3f507004 modify memory at 3f506004
one was ubuntu
dunno about the other
being a hardware guy, when I saw the error I immediately assumed an open address line or something
but it ran memtest most of the night without coming up with anything
jmkasunich: I assume something borked inside the mem. controller in the chipset
cradek was also able to get it to fail with make -j, which uses lots of memory (but nothing RTAI, RTAPI, or EMC related)
I have a machine that I may be able to test 1.5G in
but it's odd that another PC aswell
SWPadnos: how about something like this: http://cgi.ebay.de/Streamer-DLT-1-Overland-LoaderXpress-bis-800-GB-84122_W0QQitemZ8767858994QQcategoryZ79407QQrdZ1QQcmdZViewItem
another possibilty is the MMU is mapping the same physical memory to two virtual addresses
well, it shouldn't be ;)
not that that really means anything
heh, that's why we call it a bug
jmkasunich: any idea what a iommu is?
but htere are a lot of people using 1G+ machines, and I haven't seen a lot of complaints
it's the memory management unit for I/O operations (basically DMA for I/O)
using them for EMC? or just using them for vanilla linux?
* fenn boggles at using 1G for EMC
jmkasunich: I remember reading some changelogs of more recent kernels, that say something about iommu fixes
alex_joni, that library may be OK, but it would take an awfully long time to fill ;)
jmkasunich: this might prove interesting: http://kerneltrap.org/node/2450
1 gigabyte ram
alex_joni: Do you know what command I ought to use to get the emc2 sources from your site?
Ah, I'm waiting for 8GB ram to become commonplace
you mean the debian way? or the tar.gz?
ok, you need to edit /etc/apt/sources.lst
I got a command from chris the other day that setup all the dependencies.
Got that now.
and add a deb-src http://dsplab.cs.utt.ro/...
just like the deb http://dsplabs..
you have in there
ok, apt-get update
then apt-get build-dep emc2
if I remember it right
okay. That's it.
going away to try.
that should pull all packages needed to be able to compile emc2
after that: apt-get source emc2
or get it from CVS, shouldn't matter
I wonder if somebody (hint, hint) would be kind enough to put that info on the TESTING wiki page
under "other ways to get emc2 testing"
whats the diff of SFPM and FPM ?
sfpm means surface feet per minute
probably more used on lathe, to refer to the surface of the part moving past the tool
'surface' as opposed to ?
[01:52:04] <Jymmm> http://www.machinist.com/calculators/
jmkasunich: I can do that tomorrow.. need to test it as I do it, ok?
actually, SFPM or SFM is the rate the cutting edge moves over the workpiece, regardless of the cutter type (saw, endmill, etc.)
I just couldn't understand why the word 'surface' was used - read it in a couple palces too.
right - it's the tool moving over the surface
as opposed to whay, in/thru some material?
no, as opposed to e.g. feedrate
feed rate is 100, but the SFM is dependent on the tool speed and diameter
yeah, but that wouldn't be OPPOSED to feedrate, that would BE ffedrate.
as opposed to, ie, "different from"
Unit Of MEasure == IPM, FPM, etc
feedrate is usually ipm
so, ipm * 12 == FPM
and tool rate is SFM, which is dependent on spindle RPM, and the tool diameter (and a little bit on the feed rate)
err what SWPadnos said =)
ok so ipm/12==FPM, which is the same as SFPM ?
Ok, then why not just FPM instead of Surface FPM ?
unless SFPM is different from SFM
SWPadnos but is your disagreement justified or smartass?
justified, I think
I have no time for being a smartass tonight :(
same to you
the reason I say that is that the calculation is dependent on the spindle RPM and tool diameter
the feedrate is used to set the depth of cut
hold your horses.... that is the FIRST time I've heard clipload mention depth
sorry - I meant d\chipload
err - just chipload
you set depth f cut once you have feedrate and tool diameter, and it's based on the spindle horsepower
back in a minute
so what your saying/thinking is 1/4" diameter endmill 1/2" into the material would be .5"
Surface Feet Per Minute (SFPM).
Linear velocity of the thermal spray gun as it traverses the length of the workpiece.
Also, the circumferential velocity of the substrate.
ok, here's how you do it:
1) look up the "optimal" cutting speed in SFM (example, 100)
2) choose the tool you want to use, and get its radius (example, 1" tool, 0.5" radius)
no - that's next
there's a cutting speed, and there's chip thickness
3) calculate the necessary RPM to get the correct SFM with the chosen tool
[02:07:37] <Jymmm> http://www.owwm.com/Math/SFPM.asp
sure - or the other one you linked to
Jymmm: thermal spray gun tutorial isn't the best way to understand the difference between sfpm and feedrate
fenn just waht google returned
anyway, the speed of this tool is pi*r = 1.57 inches / rev, or ~0.125 SFM / RPM
with me so far?
think so, go on
ok. remember step 1, where the SFM was looked up, and it's 100
to get the required spindle RPM, divide desired by SFM/rev, and you get about 800 RPM spindle speed
... divide desired SFM by ...
OH! so SFPM is determining how fast (or how much) material I can feed thru it?
now, we take chipload into account - asy 10 mils, so the feed rate needs to be 800 RPM * 0.01 inch / tooth * number of teeth (flutes)
not quite ;)
jymm - SFPM is how fast the cutting edge moving thru the material
if you go over the recommended SFPM the tool bit will overheat.. you might not be able to cut at the recommended sfpm because of other factors
so if we have a 1", 4 flute cutter (should have mentioned that before) ;), we need 800 * 0.01 * 4 = 32 IPM feed rate
the last thing to calculate is the depth of cut, which is dependent on spindle horsepower (and also on axis motor power, and tool strength)
this is another thing to look up in the tables - the removal rate per horsepower for the material being cut
and machine rigidity..
yep, but hopefully that isn't a problem ;)
well, it usually is a problem
for just about any home shop machine it is the limiting factor
so, from the tables, I think aluminum is 6 cubic inches per horsepower *at the tool*
ok - fair enough
I bet the average Bport takes HP limited cuts no more than 20% of the time
could well be. I'd imagine that the tool strength also plays a part a lot of the time
and spindle speed
hard to take a 2HP cut with a 1/4" mill unless you spin it at 20K rpm
yeah, but anyway - I'm not sure there is any way to calculate speeds based on machine rigidity, unless you use an FEA package
speeds are by the book
feed depends on rigidity
there is a classic post on rec.crafts.metalworking on this topic, I'll find it
speed and feed should be by the book, with depth being based on rigidity and HP
because chipload is directly dependent on speed and feed
wouldn't depth have to do with tool as well?
[02:18:05] <jmkasunich> http://groups.google.com/group/rec.crafts.metalworking/browse_thread/thread/8a2270dfeaa21096/c44d2304c2664a6e?lnk=st&q=bastow+dynamite+shift+cuppa+group%3Arec.crafts.metalworking&rnum=3&hl=en#c44d2304c2664a6e
jmk is probably thinking of this http://yarchive.net/metal/empirical_feed_rates.html
gawd what a url
yes - somebody saved it, great
i read that whole list of articles top to bottom one day
Jymm: read that article
jmkasunich will do.
I'm trying to understand the whole concept, chipload was questionable as it never accounted for depth, which threw me a curve. Once I 'get it' I want to create a calculator for it (maybe with the tables, but doubtful)
Cause right now... dougal fir can kiss my ass!
read the article - it does an eloquent job of explaining why no calculator will ever replace experience and skill
jmkasunich If I write a calculator, at least I'll know if I understand it all correctly.
I can machine mdf no problem, I can machine Ironwood no problem, but toss on douglas fir and shit hits the fan! So before I do anything else I want to figure out whats goin on here.
I'm suspecting the alternating densities in the grain are a factor
douglas fir is much more anisotropic than ironwood
ie. more grain in the wood
what problems are you having specifically?
fenn: Almost every singal time I've used douglas fir, the results are mucked up in one way or another... this time, chipouts and (best I could explain) gouges in certain areas.
are you climb milling?
up helix or down?
1/2" 90 deg and 1/4" downcut
might try using a roughing endmill..
the 1/4" downcut is for peck drilling four holes, then to cutout the piece.
like this http://static.flickr.com/18/68122545_7ae2b17d59_m.jpg
that looks nice
not enough detail to see any problems though
what's the blob inside the "o"
my thumbprint in the paint
a bug landed on the piece when it was drying
mucked up in douglas --> http://static.flickr.com/31/68122546_d54054ffe1_o.jpg
interesting burn marks
or is that stain?
I clearcoat, then carve, then stain
it's to help contrast the piece
this might be a better example http://static.flickr.com/35/68122547_7d6160cf29.jpg
[03:01:24] <Jymmm> http://static.flickr.com/34/101009611_3962bf3dfe.jpg
jymmm try more cutter flutes or do a roughing pass first
what's a 'roughing pass'?
I'm using two flute downcut right now.
roughing means to cut out most of the material in the first pass then shave off the last bit on the final pass
i think the real problem is that you're using V bits and so you cant do climb milling
It's jsut strange, I can't put my finger on anything specific, but with any other species it's fine, just hates douglas! lol
the problems are never the same.
maybe the wood was cured improperly
This has been other 4 months and at least 5 different batches of wood.
I always buy extra b/c it seems to always muck up.
fenn: This turned out very nice (cause it aint douglas - LOL) http://static.flickr.com/34/101009611_3962bf3dfe.jpg?v=0
i see some wobblies around the corners and at the top/bottom of letters
and i dont see what's wrong with the ruth & tom one, except maybe depth is too low
ruth & tom is only 1/2" thick
you measured the thickness of the piece right?
actually it looks like it's bending upwards in the middle
see how the radius of the cutter is larger in the middle of the board than at the edges?
in which sign?
the stained ruth & tom
Jymmm is now known as Jymmmm
I think that one is just a fluke, especially if you compatre it to the solver version, which is what it should have been.
oh hell, just run everything type thru a translater
oh hell, just run everything type thru a translator
i think they call those spellcheckers
I just call em typos
how do you get a square internal corner? like on the end of the cross-bar of a lower-case t
i know the cam software does it, but i wanna understand how the bit moves to make that shape
stained R&T ?
it looks more square on the silver one
the t in "Ruth"
probably lifts the V tool as it comes out of the corner
so the cut gets narrower and narrower,
but it doesnt get narrower, and the bottom looks v shaped still
does it have to move up/down twice to make the two corners?
fenn hold on....
Jymmmm: if you could show me a toolpath for a lowercase "t" it would be niec
could you paste the gcode somewhere?
I have to remember what font face I used first =)
gimme a few, search for the file(s)
ah.... TNR if the font
Times New Romain
2" tall 't' http://pastebin.com/560650
* fenn was away toasting toast
* Jymmmm is being dragged happily by the gf to watch a movie!
03cradek * 10emc2/scripts/emc.in: RIP should only find configs in its own tree
jmkasunich is now known as jmk-sleep
sorry xchat is being dumb
i was playing around with japanese character set encodings and it doesnt listen to my auto-join preferences
Saw 2 is AWESOME! But if you havne't seen Saw, be sure to watch it first!
[05:47:52] <Jymmmm> http://www.sawmovie.com/
Jymmmm is now known as Jymmm
fenn were you able to extract what you wanted to know from that gcode?
what was that (specifically)?
how it did the square corners
and what is it doing?
the path has little "Y" branches at the end to cut a square corner
so it's just tapering upwards
giving the illusion of squareness
well it actually is square in 2D
you're talking the right side of the line, correct?
it actually does it on all the edges, but your depth is set wrong or something
or maybe they're slightly rounded corners
looks rounded to me =)
[05:54:25] <Jymmm> http://static.flickr.com/41/101020064_9c3aa8cdfb.jpg
well you can never make a corner sharper than the radius at the top of the V bit
I'm suspecting it's the cam that's doing that
I guess your 'square' corners are just a v tip running around the edge of a square would reproduce the same results.
yeah but not in one pass
and it wouldn't look as nice
If I had used a 1/16" 90 deg tip it would
but I was using a 1/2"
you could go over it again with the 1/16" bit after the 1/2" bit
Think about it... a 1" square and you machine on the lines with a v cutter then clear out the insides would appear like a square
but it would be fugly
you only need to touch up the corners
if you do more than that youre gonna have places where the finishing pass doesn't line up perfectly with the roughing pass
square corners are overrated anyway
oh, is THAT why you wanted the gcode =)
i wanted the gcode because i'm planning on writing some cam software
lol, ah ok
the same algorithm does square corners as well as lines of various widths
alpha1125 is now known as A-L-P-H-A
ValarQ_ is now known as ValarQ
food time.. brb
[12:33:24] <Jymmm> http://media.putfile.com/assbeating
I think I've almost got my ubuntu system working like I want (key shortcuts and such)
have not gotten into changing things around - looks very user friendly for non linux users like me.
I like being able to click on the emc2 icon ;)
and also being able to drag it to the desktop
yeah, cradek's packages make it all very convenient
and then automatic updates.. My father will love that.
do you have over 1gb to put in that thing for shits and giggles?
I actually have a 2GB "server", and right now I'm installing ubuntu inside qemu (PC emulator) specifically to test with more than 1GB RAM
I hear cradek has a test that very quickly blows up on your stem?
yes if there is over 1gb mememory - when we went back to 1gb it ran fine
though the installer (running with 900 megs RAM) just gave me a biarre error...
some package (it wouldn't tell me which one) didn't install right, it said
for emc or ubuntu?
I'm still working on getting the base install done
odd - the only issue on an older machine is the cd drive couldn't read the burned cd - at about 80% of the final install ;)
I bunrned one slower and it worked fine. It didn't go on with the install at that point.
oh - I did have one lock up at 71 percent when it was trying to configure the x-windows if I remember right. it was the old stb video card I had in it - replaced it with a matrox and it installed fully
boy over the last few days I have installed it quite a few times ;)
did you read that I had installed it on 2 differnt machines and they both didn't work with memory over 1gb?
the only thing the same was they where both pentium 4's
both worked just fine with 1gb or less
I haven't paid attention to all the details, but cradek has told me some things
the latest pattern I identified is if there's >1G *and* realtime is loaded (and even after it's loaded and removed) the system is unstable.
before realtime is loaded, the system is fine.
interesting ,-been saying that a lot recently
I'll try to reproduce it here today.
rayh suggests we fix it by saving money on ram, or putting it in a nearby windows machine that needs it
or send it to him?
no, I think he was somewhat serious - no machine dedicated to emc needs anywhere near that much ram...
so in all seriousness that's a possible solution I have in mind if something else doesn't pan out (it's a rtai bug for instance)
right - it just happend that the machine I was testing had over 1gb
I could easily make the -magma kernel ignore everything over 1G - that would let it run reliably on any machine.
can you set the memory size the real time uses?
crap, as I was afraid, my good machine has only two memory sockets
crap, as I was afraid, BOTH of my machines have only two memory sockets
so my machine is still running?
yeah, it's fine
jmk was in last night wondering what you had found. He was supriesed.
I forgot to thank him for the work he did on the base_period issue. EMC2 is running great now.
unbelievable compared to when I first started playing with it.
yeah I'm glad he found that problem.
I was able to run that computer you are playing with at a period of .00002 and axis was smooth and usable.
with 1gb in it ;)
emc2 still cvs only?
Jymmm, There is a deb for ubuntu breezy
breezy being the latest?
hey ray - could you explain your static ip address problem? That is how I setup ubuntu at work (we don't have a dhcp server so I have to give ubuntu a internal ip)
(thinking about it as I was drifting off to sleep ;)
jymmm - the ubunto with cradeks install script for emc2 is painless.
[18:10:52] <skunkworks> http://timeguy.com/cradek/emc/ubuntu
must have installed it 5 or more times on 3 differnt computers over the last few days with only the 1gb limit issue. (no problems with the installs)
Um, it there were no issues, why was it installed 5 times?
skunkworks: if it doesn't find a network interface at install time, it doesn't enable the network apt repositories. ray set up the network interface later, so he had to enable the network repositories afterward.
we where trying to figure out what was going on with one of my computer - emc2 would not run
The static IP problem was simply a failure to uncomment the package directories in /etc/apt/sources.list
it ended up being an issue with over 1gb limit
ah - when it doesn't see the dhcp server I just set it up manually at that point - so I didn't have a problem I suppose
Since the installer did not get assigned a dhcp addy it said that it wasn't connected.
[18:15:34] <cradek> https://mail.rtai.org/pipermail/rtai/2006-February/014119.html
hey look at this
(it give an option to set the ip addressses up manually during the install
Nope. Just go into the sources list and fix em by hand or with synaptic.
Then when you are connected to the outside world, issue apt-get update and you should see all the stuff you need.
cradek: Whats the ogo stuff to mit for?
Yes. That is how I assigned my box to my net but it still didn't fix the apt stuff.
what I am saying is that durning my install - it didn't see the dhpc server so it let me set up the ip address/gateway and dns server manually. I didn't have issues with that.
cradek : gpg --keyserver pgpkeys.mit.edu --recv-key BC92B87F
Jymmm: I don't understand the question
cradek Didn't you write this --> http://timeguy.com/cradek-files/emc/ubuntu/emc2-install.sh
cradek: Why does it contain pgp/gpg stuff?
but I can't parse your question "whats the ogo stuff to mit for?"
because the apt repository is signed by me, so you know you're getting valid packages
cradek, I was looking at the gnome Yelp help stuff.
cradek ah, ok =)
that key BC92... is my public key
rayh: what's that?
cradek: np, just wasn't aware you were signing is all
It's the program behind the help icon on the top bar.
rayh: what are your thoughts?
The ubuntu FAQ Guide that displays there is a complete "docbook" book.
XML of course but we can create that from LyX.
This might give use the chance in emc2.2 to add emc, hal, cl documentation without any additional programs.
And not a lot of additional writing.
do you think it's better than using html?
I'd be really hesitant if it means we are tied to having a particular desktop environment in order to read our help
The advantage with Yelp is that you get a tree display along with the linked text and images.
I agree the tree is nice
hypertext, images, etc. come with html, but a tree is harder
If all we have to do is export a lyx file as docbook it would make nice docs for ubuntu easy.
I'll play with it when I get the chance.
our lyx files are in outline format (1.1 1.2 1.2.1 etc)?
if that would turn into a tree it'd be very slick
It will also display man (troff)
Kinda close - has your idea for a fix.
I made a halcmd.1.gz and stuffed it in /usr/share/man/man1
and Yelp displays it.
along with all the other man stuff.
you should stick the man page in cvs, and we'll get it into the distribution
It is in the distro in emc2/docs
Could we automate that with the make install?
cool, I didn't know we had that
I'm working on rtai right now though
don't let me forget
Yelp even lists it just like any other.
np I'll bug you from time to time.
I suspect that jmk wrote it by hand in troff format.
If anyone cares --> http://www.harborfreight.com/cpi/ctaf/Displayitem.taf?itemnumber=36196
Jymmm, how are you pear.php?
03rayhenry * 10emc2/docs/man/man1/halcmd.1: Added EMC to title to clarify withLinux HAL
A-L-P-H-A: *I* don't use/trust pear personally.
trust, why not?
A-L-P-H-A: none of the code is audited
I'm trying to learn db_objects, for rapid form generation.
Jymmm, so make your own login script, and only allow trusted users.
A-L-P-H-A heh, theres more to it than JUST login
argh. fuck it.
I just want to learn how to use db_object so I can use the quick form generator.
A-L-P-H-A If I knew and trust the person who write it, that's another story.
Borland / Corel's database
load up a database, hit F8, and get a nice form, including chile tables
A-L-P-H-A db_object doesn't even show up in a search
no. it's full name is.
[20:02:42] <A-L-P-H-A> http://pear.php.net/manual/en/package.database.db-dataobject.php
package I'm trying to use is. http://pear.php.net/manual/en/package.database.db-dataobject-formbuilder.php
A-L-P-H-A and you have all this? http://opensource.21st.de/18.html
I'm still on the issue of setting up the ... <sec>
I'm still on the issue of setting up the ... DSN....
[20:07:11] <A-L-P-H-A> http://pear.php.net/manual/en/package.database.db-dataobject.intro-configuration.php
documentation here SUCKS ass.
[20:07:19] <Jymmm> http://opensource.21st.de/16.html
Jymmm, I'm using it as reference now.
I don't have an proper INI file setup.
lol, click on 'generating forms'
which is what I'm having troubles with. :)
A-L-P-H-A: I dun know man... I think you would have better luck removing and replacing you eyeballs than mucking with pear.
Okay... I finally can connet to the DB via PEAR
jmk-sleep is now known as jmkasunich
jmkasunich: Read the old posting last night =)
jmkasunich: My 'chipload calculator' was more of an applied understanding of speeds and feeds than an absolute environment thing =)
03cradek * 10emc2/tcl/tkemc.tcl: use a nice truetype font instead of scaled bitmaps.
03jmkasunich * 10emc2/src/module_helper/Submakefile: MODULE_EXT needs to be defined unconditionally
Hi - whats up?
I found reference to a problem with 1G that was fixed in rtai-3.3
rtai-3.3 - what version does emc2 use?
so I built a new rtai-3.3 package but that does not fix your machine.
skunkworks: all of them ;)
it was rtai-3.3-test4, a 3.3 prerelease.
skunkworks: I've run rtai-24.1.11, 24.1.12, 24.1.13, 3.0, 3.0-test1, 3.1 and 3.2 (and now on ubuntu the 3.3?)
but that's in 2 years
I didn't like it at all ;)
that link that you sent made it sound like 1gb was a problem - not more or less. maybe I read that wrong
they all use different schemes to compile
it does say that.
I thought it might be related anyway.
I'm not sure we've tested enough to be positive what our pattern is.
you where trying to get 1gb in your machine - not posible with 2 slots?
I was trying to get 1.5, no not possible
I can get 1
cradek go for 2gb =)
I can't, I only have 512 modules
jepler I think was also going to try it on one of his machines that has like 2gb
I don't think it was me ;).. it did it on 2 machines with isn't the greatest test.
03jmkasunich * 10emc2/docs/NEWS: NEWS update
jmk - forgot to thank you last night - The fix you did to the period (going by clock cycles) really seems to have sped emc2 up. thank you
of course, it was me who added the original timing code that slowed it down, so..... ;-)
it worked fine for a while ;)
skunkworks: it worked fine 80% of the time
yeah, the timing code was disabled for a long time
cradek: The requested URL /cradek-files/truetype/ttt_2.0.tar.gz was not found on this server.
but sometimes it worked very bad ;)
was the full url
then cradek asked about detection of RT overruns, and I started trying to get the timing stuff to work
oh sure it's my fault
LawrenceG: oops, checking
yep, its cradek's fault
jmkasunich: without remembering why you abandoned it last time ;)
you seem to have got it - I was running a period of .00002 and it was very usable.
cradek: np... deb works ok
* alex_joni needs to try this now..
last time I ran it I was at 0.000025 without the speed improvement ;)
LawrenceG: _ vs -
alex - on how fast of a machine
Centrino 1.4Ghz, running at 800Mhz for some strange reason ;)
probably because ACPI is not enabled to tell the processor what speed it should use
ok - I have to go again. working on the basement. will be back later. cradek - so you think it is a rtai problem?
we all do sam ;)
skunkworks: yes, but I wish I could do more extensive testing to be sure.
like people have mentioned (ray) that running >1gb for an emc machine is odd. But people will.
the "fix" might be to ignore ram over 1G (it's a kernel config option)
cradek: wonder what changing the 1/3 G to something like 2/2G would do
alex_joni: I don't know what you mean
well, as far as I understand it linux mem.management works like this:
there is a 1G address space for system, and 3G for user stuff
otoh if you have physical ram more than 1G the stuff outside will get mapped into system address space first, and used afterwards
this probably works ok for the kernel & everything above it, but rtai might not have a clue about those address changes
gosh I hope you can really make something out from my poor explanation
cradek - some time next week if you want to get into my workstation that is doing the same thing we could.
03cradek * 10emc2/src/Makefile:
gene had a problem on bdi where sudo empties $PWD at the sudo make setuid
step of rip build. Maybe this will fix it.
Jymmm, figured it out.
03jmkasunich * 10emc2/TODO: added to TODO
A-L-P-H-A what was it?
it was wrong
When you do rough/finish pass... is it typical to use a different tool?
or just to alter the feed/speed?
Jymmm: often you use a smaller tool for finish
or more flutes?
Jymmm, oh the whole... I figured it out... well the basics.
A-L-P-H-A Ah, ok.
not the advanced stuff, like foreign key groking.
which I'm about to sign up for.
03cradek * 10emc2/debian/control: new vti stuff requires this
I was asking because I'm doin a lot of v-carving, so using a different tool doesn't always apply.
but fenn mentioned somethign about it last night.
sign up for a newsletter thing
03alex_joni * 10emc2/src/hal/drivers/ (hal_vti.c hal_vti.h): intial comit for the VTI driver written by Eric H. Johnson, I ran the files through dos2unix, and started to fix some compile errors, lots more to come.
03alex_joni * 10emc2/src/Makefile.inc.in: adde VTI to the list of possible drivers. for now set to 'n' because it doesn't compile, and it will break the whole build
03alex_joni * 10emc2/src/Makefile: added VTI driver to the Makefile. to enable it, edit Makefile.inc.in
I'm on sentimental idiot. itunes is on Random songs... and it's hitting songs that I used to listen with my exGFs... kinda an odd feeling.
time to block those it sounds