#emc-devel | Logs for 2008-05-16

[00:48:07] <rayh> odd. I see an fr version of HAL users manual but not an en version in the 8.04 release.
[00:49:09] <rayh> or am I missing something more than the ability to see.
[00:50:27] <SWPadnos> err - where are you looking?
[00:59:50] <rayh> /usr/share/doc/emc2
[01:04:48] <rayh> on a system installed from the emc2 package.
[01:22:34] <jepler> rayh: I'm not sure why HAL_User_Manual_fr.pdf is in there -- all the hal specific stuff should be in the EMC2_Integrator_Manual.pdf, I think
[01:23:19] <rayh> I was looking to compare released versions of it with what I'd compiled here but it just ain't in there.
[01:24:18] <jepler> for instance, EMC2_Integrator_Manual.pdf has HAL Tutorial, Canonical Device Interfaces (in section 'hal specifics') and Hardware drivers (in section 'EMC related HAL')
[01:24:20] <cradek> fwiw, in my dapper 2.2.5 emc2-sim I have both english and french pdf files
[01:24:43] <jepler> jepler@hardly:/usr/share/doc/emc2$ ls *.pdf
[01:24:44] <jepler> EMC2_Integrator_Manual_fr.pdf EMC2_User_Manual_fr.pdf HAL_User_Manual_fr.pdf
[01:24:47] <jepler> EMC2_Integrator_Manual.pdf EMC2_User_Manual.pdf
[01:24:56] <rayh> When I build 2.2.5 here it produces a copy.
[01:25:07] <jepler> rayh: yes it builds files that aren't used, because they contain duplication
[01:25:10] <cradek> oh I see
[01:25:41] <jepler> you'd want HAL_User_Manual.pdf as a separate document if you were going to use hal separate from emc, but for emc users we just put everything in the integrator manual
[01:25:50] <rayh> There is quite a bit of overlap between these.
[01:26:20] <rayh> user, integrator, hal
[01:26:41] <rayh> I'm just stumbling across things while getting ready for the classes.
[01:27:14] <rayh> Thanks for looking.
[01:27:37] <SWPadnos> it's tough to decide whether a big manual with configuration through G-code is better than separate mannuals for HAL, integration, development, operation, troubleshooting, classicladder ...
[01:28:33] <SWPadnos> do the lyx-to-pdf tools support something like a master document with links to other docs (in other files)?
[01:28:41] <jepler> SWPadnos: yes that's exactly what it does
[01:28:55] <jepler> well, maybe I don't understand what you mean
[01:29:02] <SWPadnos> hmmm. I thought you ended up with a big single file with all the sub-documents included
[01:29:04] <jepler> there are several .lyx files which include other .lyx files
[01:29:10] <jepler> are you asking for links between pdfs?
[01:29:14] <SWPadnos> yes
[01:29:27] <jepler> pdfs are for killing trees. html files have nice links between them, though, and that's what they were made for
[01:29:38] <SWPadnos> some docs I've gotten have a master index, which then point to other files for the detail information
[01:29:46] <SWPadnos> odf dics
[01:29:48] <SWPadnos> err
[01:29:50] <SWPadnos> pdf docs
[01:30:33] <rayh> SWPadnos, We make the big doc using a master Lyx that includes the other files
[01:30:52] <SWPadnos> right - I'm thinking of something different :)
[01:31:08] <jepler> pdf files are for killing trees. html files are for hyperlinks.
[01:31:18] <jepler> if you want something that's linked out the wazoo, use html
[01:31:29] <SWPadnos> it's not about hyperlinks between files, it's about organization and partitioning
[01:31:43] <rayh> But still one big doc?
[01:31:46] <SWPadnos> I think my electronic CAD program does this - there's a master document which is fairly small, but will cause other files to be loaded
[01:31:49] <SWPadnos> no, separate files
[01:32:04] <SWPadnos> so I could load the HAL Integrator manual if I know that's what I need
[01:32:09] <rayh> When the end user views it, what does it look like.
[01:32:27] <SWPadnos> or I can load the "top level" file, which will have short descriptions of the other files, along with links to them
[01:32:55] <SWPadnos> separate files, but one essentially lists the others, and lets you easily load them by clicking link-like things
[01:33:15] <jepler> SWPadnos: I believe that's important; that's why I improved the html export, which makes things like that work splendidly.
[01:33:18] <rayh> Lots of our chapters are individual LyX files
[01:33:26] <SWPadnos> adobe reader also has search options, so you could load "the emc docs", and search all the files automatically
[01:33:30] <SWPadnos> heh
[01:33:39] <SWPadnos> do users get HTML with an installed system?
[01:34:13] <jepler> for historical reasons they get pdf
[01:34:13] <rayh> Only if they build.
[01:34:20] <SWPadnos> I'm not sure if all that hierarchical stuff is supported by the free viewers anyway (especially evince, which seems to be the least useful of them)
[01:34:20] <jepler> with 2.3 I intend to switch to html
[01:34:24] <SWPadnos> ok
[01:35:00] <SWPadnos> just thinking out loud - I have no complaints
[01:35:28] <rayh> I'm still thinking that docs and control ought to be separate packages.
[01:35:34] <jepler> at one point there was a "what doc format do you want" poll but now I can't find it to see what the results were
[01:36:13] <rayh> With packages of docs in your favorite language.
[01:36:21] <rayh> and perhaps in your favorite format.
[01:36:46] <SWPadnos> I like one big file because it's easy to search. then again, I also like smaller files that are more tailored to a specific subsystem or task :)
[01:37:13] <jmkasunich> if they are files, they should be HTMP
[01:37:14] <jmkasunich> L
[01:37:16] <SWPadnos> so smaller files that are also easy to search would be the best of both worlds (unless there are 1000 files ;) )
[01:37:17] <jepler> SWPadnos: if only I'd been able to figure out the gnome help system, it would have provided a full-text search of the html documentation. unfortunately, it bested me.
[01:37:18] <jmkasunich> pdf is for printing
[01:37:38] <jmkasunich> * jmkasunich channels jepler
[01:37:38] <SWPadnos> yeah - maybe several more heads will be able to help you get it :)
[01:37:59] <rayh> jepler, I tried for a few days and it beat me also.
[01:38:28] <jmkasunich> I need to learn never to cook anything ambitious on a weekday - if I'm lucky I'll be eating dinner by 10pm
[01:38:41] <jepler> (fwiw the pdf docs are 10 megs and the html docs are 6 megs. that's including both languages and all the same images as far as I know)
[01:38:46] <rayh> whatchamaking?
[01:38:49] <SWPadnos> make 23-hour stew in a crock pot
[01:38:55] <jmkasunich> chicken caccitore
[01:39:02] <jmkasunich> or however you spell that
[01:39:02] <rayh> Ah sounds great.
[01:39:11] <jmkasunich> big pot of it, most will go in the freezer
[01:39:20] <jmkasunich> so the next 5 times it will be easy and quick
[01:40:48] <rayh> With the advent of my fast link, most of my objections to 10 Meg have gone away.
[01:41:14] <jepler> funny how that is
[01:42:36] <rayh> But if we get 30 languages...
[01:42:53] <jepler> that would be one of my dearest fantasies
[01:42:56] <rayh> That will really dwarf the control software
[01:43:31] <jepler> I'm surprised to see that even the most trivial .ko are about 88KB
[01:44:11] <SWPadnos> this makes it look straightforward: http://scrollkeeper.sourceforge.net/documentation/scrollkeeper_example2_manual/ar01s02.html
[01:44:20] <SWPadnos> so they must be wrong :)
[01:44:26] <jepler> SWPadnos: I probably just overlooked something
[01:44:44] <SWPadnos> it's XML - easy to overlook everything
[01:48:59] <jepler> well -- here's my broken attempt at omf support. you really can't test it short of building a deb and installing it. patch is against TRUNK. http://emergent.unpy.net/index.cgi-files/sandbox/broken-omf-support.patch
[01:49:09] <jepler> bbl
[01:49:18] <SWPadnos> see you
[01:49:37] <rayh> Have a good evening guys.
[01:52:22] <SWPadnos> later ray
[01:53:20] <cradek> I wonder what I should be doing to get ready for fest.
[01:53:40] <cradek> I put my list of stuff on the wiki. I need to install the machines I'm taking
[01:54:59] <jmkasunich> I've been debating if I should bring my main machine, or leave it here and just bring my older one
[01:55:07] <SWPadnos> the shoptask?
[01:55:13] <jmkasunich> no, computers
[01:55:16] <cradek> I think I'm going to take three (!) laptops
[01:55:18] <SWPadnos> oh, phew!
[01:55:31] <cradek> one to run each machine plus my normal one
[01:55:36] <jmkasunich> the main PC has the compile farm and serves my website - the farm can run at Rolands, but the site can't
[01:56:26] <SWPadnos> can you ssh into it when it's at home?
[01:56:45] <SWPadnos> oh hmm - do you have a hardy VM yet?
[01:56:45] <jmkasunich> yes, as long as nothing goes wrong
[01:56:52] <jmkasunich> lno
[01:56:54] <jmkasunich> no
[01:57:07] <jmkasunich> I hear beeping, I think its food
[01:57:11] <jmkasunich> bbl
[01:57:14] <SWPadnos> see you
[01:57:19] <SWPadnos> 10:00, right on time
[01:57:24] <cradek> funny
[02:34:36] <jmkasunich> damn that was good - worth the wait
[02:58:18] <cradek> jmkasunich: what do you hope to get done at fest? or do you know it's useless to plan?
[03:14:34] <jmkasunich> its pretty useless
[03:14:49] <jmkasunich> gotta try to do axis-joints stuff mostly
[03:16:46] <cradek> the thing I want to be sure to do is the rigid tapping time lag comp. I don't have the hardware, and I don't think there's any sure way to simulate it
[03:17:17] <cradek> it will take some playing to fix, at the least
[03:17:27] <jmkasunich> how will you measure success? looking for the part to get pulled up out of the vise?
[03:17:38] <cradek> with the acme rod you're bringing
[03:17:46] <jmkasunich> oh, thats right
[03:18:10] <jmkasunich> (I do have that written down)
[13:04:05] <SWPadnos> oh, now that's an interesting idea - prevent auto mode until homed
[13:09:29] <alex_joni> SWPadnos: that's an old idea ;)
[13:09:39] <SWPadnos> it's still interesting ;)
[13:09:46] <alex_joni> right.. didn't say it wasn't
[13:10:04] <alex_joni> but I'd find it more interesting to have that configuration dependable
[13:10:17] <SWPadnos> yes - of course I thought about that :)
[13:10:56] <alex_joni> and we get back to Runlevels..
[13:10:57] <SWPadnos> essentially have inputs (or messages ...) that must be on to allow going into that mode
[13:11:00] <SWPadnos> yep
[13:11:05] <SWPadnos> and transitions
[13:11:22] <alex_joni> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?RunLevels
[13:11:49] <SWPadnos> yep - I remember some of those conversations
[13:25:35] <rayh> I've never known a shop guy to try and run a program without homing. The routine is power up, home, go to work.
[13:26:16] <rayh> I don't even know if you could switch to running a program without homing on a Mazak.
[13:26:35] <SWPadnos> that's what the email suggested
[13:26:41] <rayh> It would never enter the head of an operator to try.
[13:27:08] <rayh> That's not what the email suggested.
[13:27:10] <SWPadnos> but at the moment, I don't think you can configure EMC to do that
[13:27:33] <rayh> "Don't let me is what the email suggested."
[13:27:58] <SWPadnos> oh - I was responding to your musing that "I don't even know if you could switch to running a program without homing on a Mazak."
[13:28:11] <SWPadnos> that implies that the Mazak might not let you ...
[13:28:14] <rayh> I understand.
[13:28:27] <SWPadnos> (I'm assuming before the EMC conversion ;) )
[13:28:37] <rayh> I wonder if my books even say.
[13:32:25] <SWPadnos> is it just me and my mailer, or does it seem that "reply-to" is picking up both the list and the sender now?
[13:32:45] <SWPadnos> I wonder if I managed to set some "reply to all all the time" option
[13:34:06] <rayh> Don't see any reference to no auto unless homed.
[13:34:33] <rayh> I think such a thing could be done with halui and either cl or a comp.
[13:35:12] <SWPadnos> I'm not sure. UI->task is a direct shot
[13:35:25] <SWPadnos> halui can only add to that, not step in between
[13:36:06] <SWPadnos> it might be possible to have a watchdog that looks at "homed" for all axes, and sets machine off (or jog mode) if it also sees that EMC is in auto mode
[13:36:14] <rayh> Would it matter if it stepped between or reversed the status.
[13:36:18] <SWPadnos> but that's reactive
[13:37:00] <SWPadnos> hmmm. I'm not sure there's a "has-homed" HAL pin
[13:37:01] <rayh> If -- not all homed -- and status auto -- set mode manual.
[13:37:04] <SWPadnos> (for each joint)
[13:37:25] <cradek> on the BOSS you can't even jog before you home, and it sometimes causes me pain
[13:37:50] <SWPadnos> interesting
[13:38:15] <rayh> That doesn't sound right at all. If a Mazak is shut down away from home you can jog it near and then press home.
[13:38:31] <cradek> if you put your thumb on a limit switch, you can jog with the wheel. this is probably fairly unsafe but I've had to do it.
[13:39:22] <rayh> We should rename the halui halvcp config in the next release.
[13:40:17] <rayh> Yea. Halui has joint.x.is-homed pins.
[13:40:19] <cradek> we should probably delete halvcp like we threatened to do for 2.2
[13:40:31] <rayh> That's it.
[13:42:06] <rayh> I seem to remember a Fanuc that would only let you jog with the wheel before an axis was homed.
[13:42:31] <rayh> Seems like Roltek's mill did that.
[13:46:19] <cradek> that seems reasonable to me, but I'd hate to put something so specific in emc - many people don't even have wheels
[13:47:01] <SWPadnos> well, there's auto, manual, and MDI - you'd want a way of preventing entering each mode
[13:47:14] <SWPadnos> manual includes jogging and homing, right?
[13:47:50] <cradek> yes
[13:48:41] <SWPadnos> so the root problem is that there's no configurable way (that I know of anyway) for an integrator to decide what modes are allowable at any given time - those are hard-coded into task (or somewhere)
[13:50:13] <SWPadnos> ok, seems jmk has the same email response problem as me. I wonder if SF changed the way they write headers
[13:50:26] <cradek> I think that's an oversimplification of the problem. locking out some of our three modes auto, manual, MDI doesn't give "jog only with the wheel, and only until off the limit switches, then require a home" for instance
[13:50:31] <rayh> I don't mind reactive.
[13:50:41] <cradek> SWPadnos: what problem?
[13:51:04] <SWPadnos> I noticed today that if I reply to a list email, it replies to the sender and the list
[13:51:09] <SWPadnos> which should be reply-all
[13:51:37] <SWPadnos> cradek, actually, it can do all that, with ladder
[13:51:40] <cradek> there's a Reply-To header in some of the list emails. It would be entirely reasonable for that to be an option (somewhere)
[13:52:02] <SWPadnos> you can enable/disable the wheel in HAL, you can do things based on limit switches ...
[13:52:24] <SWPadnos> it can't prevent key jog while allowing wheel jog though
[13:52:31] <SWPadnos> (I think)
[13:52:46] <rayh> No difference here between reply and reply to all.
[13:52:51] <cradek> yeah that kind of thing is why I said that
[13:53:12] <cradek> a basic reply should send to what's in the Reply-To header if it's present.
[13:57:32] <rayh> Looks like some users include themselves in the reply to and that gets passed through sf to the list.
[13:57:53] <cradek> yes I think it might be a subscriber option/preference - I'm trying to confirm that
[13:57:58] <SWPadnos> I'm not seeing that actually, but it could be only certain users
[13:58:51] <rayh> I don't see it in your post but do in Matt's.
[13:59:45] <SWPadnos> weird. looking at Alex's message - the official 8.04 announcement), I see from Alex, reply-to the list, to the list
[13:59:50] <cradek> I don't see the preference. maybe it's turned on for people who get the digest.
[14:00:11] <SWPadnos> same for the one from Organic Engines (subject "still crashing")
[14:00:24] <cradek> either way I'm pretty sure this is a feature, not a problem
[14:00:28] <SWPadnos> but if I reply to Alex, it goes to the list only. to organic engines it goes to both
[14:01:17] <rayh> Couldn't that be just the sender's preference in his own email program?
[14:01:21] <SWPadnos> yeah - I wasn't sure if it was a feature - like you'd use it if you're not subscribed or something
[14:01:32] <SWPadnos> nope, those preferences aren't transmitted
[14:01:53] <cradek> it's true anyone can add a Reply-To header to their emails. I'm not sure whether sf would pass it through.
[14:02:19] <SWPadnos> anyway - if you guys also get both recipients for Organic Engines and Matt, but not for Alex and Stuart, then I won't worry about it
[14:02:39] <cradek> SWPadnos: of course we do, it's just a header
[14:02:44] <cradek> no magic here
[14:02:54] <rayh> Yep that's what I get here.
[14:02:56] <SWPadnos> I examined the headers, and didn't see any difference though
[14:03:03] <cradek> SWPadnos: look again at Reply-To:
[14:03:21] <SWPadnos> maybe my mailer is hiding some header info - like don't put it in treply-to if it's the from address
[14:03:56] <cradek> if so don't tell me, since I'll go on a rant about crappy mailreaders again
[14:04:01] <SWPadnos> heh
[14:04:17] <SWPadnos> that was my confusion - I looked at reply-to and only saw one address :)
[14:10:07] <cradek> Should any existing Reply-To: header found in the original message be stripped? If so, this will be done regardless of whether an explict Reply-To: header is added by Mailman or not. NO
[14:10:53] <SWPadnos> oh - it's not problem. I see (when I "view message source" that both addresses are in the reply-to
[14:10:57] <SWPadnos> not a problems
[14:11:00] <SWPadnos> argh
[14:11:06] <cradek> this is the only related option I see
[14:12:38] <cradek> so the way it's set, I think mailman adds the list address to an existing reply-to header. I bet some people have them existing.
[14:13:01] <cradek> we can also have it remove the existing one first, so all messages have reply-to: list address only
[14:13:23] <SWPadnos> can non-subscribers send messages to the list?
[14:13:26] <cradek> no
[14:13:44] <rayh> I've been caught by that recently.
[14:13:54] <rayh> With two mail accounts
[14:13:56] <SWPadnos> ok, then removing reply-to is probably OK, though it's possible that some people want replies to more than one address
[14:14:13] <cradek> I'm not saying we should, I'm saying that we could
[14:14:14] <SWPadnos> then again, they can sign up with both addresses
[14:14:27] <cradek> if matt wants his responses separately, why would we forbid it
[14:14:42] <cradek> that's extremely useful for someone who gets digests but occasionally wants to participate in a thread
[14:14:50] <SWPadnos> right - I see no reason to remove it - people will figure it out when they get two copies of every message ;)
[14:34:08] <alex_joni> SWPadnos: fwiw, I don't have Reply-to address set in my email client
[14:35:10] <SWPadnos> right - I see that my mailer was just omitting the reply-to (if it's the same as from) in the "headers view" - when I look at message source, I see that it's doing what it should
[14:47:39] <SWPadnos> hahaha - I just got an email with "proce alerts" for first class tickets. Only $12621 for first class from Burlington to Quebec city
[14:47:48] <SWPadnos> or a 4-hour drive :)
[14:53:14] <jepler> I really am having trouble believing you didn't put an extra digit in that price
[14:53:23] <SWPadnos> I didn't :)
[14:53:32] <SWPadnos> they may have though
[14:55:04] <SWPadnos> it's possible to get tickets for only $2600 or so, so they probably have a 1 prepended. it's still funny that one could spend a few $k, and have a 9-hour travel time, instead of driving for about 4 hours
[14:55:11] <SWPadnos> even in the jeep, it would be less expensive :)
[15:50:53] <jepler> ...aaaaaand my 64-bit + rtai machine crashed again
[15:55:41] <cradek> well that bites
[19:04:21] <rayh> Mine's still up.
[19:05:33] <jepler> how much uptime?
[20:47:02] <rayh> Oh. I rebooted with a package upgrade yesterday. So I guess I don't really count that much.
[20:52:37] <jepler> that's true; as long as ubuntu comes out with security fixes frequently enough, we don't have to worry about uptimes that don't get past 21 days
[20:59:15] <SWPadnos> that only works for net-connected PCs
[20:59:40] <alex_joni> crontab -e -> reboot
[21:00:03] <SWPadnos> is that the "Windows work-alike" function? :)
[21:00:12] <SWPadnos> "reboot every once in a while"
[21:00:44] <alex_joni> no.. windows is do some weird things every once in a while