#emc-devel | Logs for 2009-06-02

[01:25:39] <mozmck> jepler: where should one send config files for some hardware?
[01:48:21] <jmkasunich> re: the big-box comments - I used to do that (and am probably guilty of some of the ones in EMC2), but I don't anymore
[01:48:49] <jmkasunich> I sometimes use a single line of /************************** something ************************/ to visually set off a section of the code
[01:49:20] <jmkasunich> for example, between declarations, typdefs, and other "top of file" stuff and the code itself
[01:52:53] <LawrenceG> big block comments worked well in fortran card decks!
[01:53:12] <LawrenceG> until you dropped your deck
[02:23:00] <mozmck> I finally got the go-ahead to release the config files I made for the CandCNC breakout board.
[02:23:23] <cradek> are you moses from fest?
[02:23:28] <mozmck> yep!
[02:23:39] <cradek> hey how was the rest of your trip?
[02:23:49] <mozmck> it was good to meet everyone!
[02:24:01] <cradek> yes that was fun. some new faces and some old.
[02:24:02] <mozmck> good. Just got back yesterday.
[02:24:19] <cradek> today was my first day back at work. it was very odd to be on a schedule again.
[02:24:25] <mozmck> Only had one breakdown. Vacuum pump went out very badly.
[02:24:33] <mozmck> :-) Same here.
[02:24:35] <cradek> vacuum?
[02:25:42] <mozmck> It's a diesel. They don't have manifold vacuum, so they have a pump to operate the brake booster and air flaps etc.
[02:25:55] <cradek> oh, that sounds important then.
[02:26:03] <cradek> no power brakes is bad.
[02:26:11] <mozmck> With the serpentine belt, when one thing goes nothing turns.
[02:26:21] <mozmck> so no water pump either
[02:26:36] <cradek> oh it siezed??
[02:27:24] <mozmck> First it dump the backside of the pump and all the oil, then about 15 miles later is seized. By then I was closer to town though.
[02:28:15] <cradek> yikes. well glad you made it anyway.
[02:28:17] <mozmck> Got towed to O'Reilly auto parts who had a pump, and I had it changed out in about 30 minutes.
[02:28:29] <mozmck> thanks!
[02:28:50] <cradek> it's such a good idea to carry tools... I'd never drive without basic tools (and jumper cables)
[02:29:16] <mozmck> yeah. I had three ammo boxes full (they make great tool boxes).
[02:29:39] <mozmck> The only thing I didn't have was the pulley remover/installer.
[02:30:23] <cradek> yeah that's a tall order for an emergency tool kit. I know where it is on the shelf at home, but that's where it would be
[02:30:59] <mozmck> I bugged Tom again today and he agreed to let me send in the files I made for our breakout board with the muxed inputs and serial outputs.
[02:31:35] <cradek> oh I didn't hear about this board - do you have a url?
[02:31:39] <mozmck> jepler actually wrote the one for the inputs a good while back, but I'll send it because he probably lost it
[02:31:58] <mozmck> I mentioned it at the fest. Let me find a link...
[02:34:16] <mozmck> http://www.candcnc.com/UBOB.htm
[02:37:32] <cradek> looks like a clever setup. it is so configurable how did you decide what to put in the sample config? do you have a full setup for all the table switches etc?
[02:37:40] <mozmck> Tom doesn't want (or have time) to support our stuff on Emc2. I told him he could just direct people with emc questions to the mailing list and site and I can help answer questions.
[02:38:11] <mozmck> I have a lot of stuff in the config, but not all yet. I need to work on it some more.
[02:38:39] <cradek> hm, that puts you guys at a huge disadvantage compared to mesa etc., who gets our full recommendation because of their great support and cooperation with the EMC community
[02:38:41] <mozmck> I did the config for my router table here at home.
[02:38:52] <cradek> (I know you already know this)
[02:39:04] <mozmck> yeah, I know.
[02:39:32] <mozmck> Tom already spends way too much time on the phone with Mach users.
[02:39:38] <cradek> a good sample config in the emc2 distribution, combined with good documentation that clearly says what to plug where and how, will do a LOT to alleviate your support burden
[02:39:54] <cradek> even if that means you recommend a combination of 3 products that plug together
[02:40:27] <cradek> the place maybe you can win is plug-and-play-ness. mesa does not have that - way too much configurability for it
[02:41:23] <cradek> maybe you can get the intermediate machines that are complex enough to have full switches and homing, but not high-end enough to need hardware stepgen etc, like mesa has.
[02:41:50] <cradek> (just brainstorming here - I don't want you guys to tell you how to do your business)
[02:41:59] <mozmck> Yeah, most of our customers build plasma or router tables.
[02:42:04] <cradek> err, "want to tell you guys how to"
[02:42:33] <mozmck> no problem!
[02:43:14] <mozmck> the modular approach seems to work pretty well for us.
[02:43:21] <cradek> but I also am wary of you telling your customers that they can expect support from the emc community after giving money to a company that says they don't have time to interact with the community
[02:43:41] <mozmck> I think that's the same thing Tom is thinking.
[02:43:59] <cradek> yeah I'm sure he sees how that puts everyone in an uncomfortbale spot...
[02:44:44] <mozmck> that is I think one reason he didn't want me to put out the stuff I have before.
[02:45:22] <mozmck> I plan on using emc and would help as much as I can.
[02:45:49] <cradek> well here's what I can offer: it's our policy to put sample configs from commercial vendors in the official packages if the vendor (or you) say you will do the ongoing maintenance and at least basic support
[02:46:18] <cradek> configs seem to always need a tune-up when we change major versions
[02:46:35] <mozmck> Tom will support his hardware but I think that he would just not be able to help users much with setting up emc with it.
[02:46:36] <cradek> if you are using the hardware yourself that will come about pretty naturally I think
[02:46:59] <mozmck> yes. I am running 2.2.8 (i think) right now.
[02:47:30] <mozmck> I need to upgrade but I have to run 200 feet or so of network cable to the shop to do that very well.
[02:47:54] <mozmck> Maybe I could just get packages on a thumb drive for now...
[02:47:55] <cradek> heh, you can do it offline (just download the deb), but online is a bit easier
[02:48:41] <cradek> unlike the xylotex (etc) it looks like your cards do need uncommon special setup (the muxing especially)
[02:49:05] <cradek> in my opinion, "supporting his hardware" means you would give the user all needed information to run the hardware
[02:49:12] <cradek> ... and I do not see that on the website
[02:49:46] <cradek> in fact there is virtually no information about how to use the hardware (except install some mach plugin thing)
[02:50:02] <mozmck> if you go the the main site http://www.candcnc.com/
[02:50:14] <mozmck> there is a "manuals" link on the right side.
[02:50:45] <cradek> ah ok, downloading the UBOB manual
[02:51:31] <mozmck> I wrote a comp file to do the serial outputs, which is what the Mach plugin does (which I wrote as well)
[02:54:49] <cradek> ok, I scanned it :-)
[02:55:21] <cradek> lots and lots of wiring options, no information about the protocol to communicate with it
[02:55:53] <cradek> but your comp will have that
[02:56:00] <mozmck> That is in the comp files...
[02:56:03] <mozmck> yes
[02:57:26] <cradek> maybe you'd like to put the files on the wiki so we can all have a look? also if you got a few customers using them, it would make us more confident about putting them in the distribution (and you would have a feel for how much support you were "volunteering" for!)
[02:57:53] <mozmck> that's an idea. how do I put them on the wiki?
[02:58:07] <mozmck> do I need a password or something?
[02:58:24] <cradek> yes, the instructions (and password) are on http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?BasicSteps
[02:58:40] <cradek> the password is only an anti-spam measure, not anti-content
[02:59:25] <mozmck> ok.
[03:11:17] <mozmck> hmmm, I've never edited a wiki. How do I create a new page? Where is it linked from?
[03:11:48] <cradek> you can make a link anywhere, and then follow it. following a link to a nonexistant page lets you create it
[03:12:03] <mozmck> oh. ok.
[03:56:57] <mozmck> I put the information here: http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?ContributedComponents
[03:57:19] <mozmck> let me know if there is a better place...
[06:22:15] <micges> jepler: in that patch there was remove #ifdef and #if 1 and #endif
[07:14:00] <micges> what this function do: void EMC_TRAJ_STAT::update(CMS * cms) in emc.cc ?
[07:14:47] <micges> is this a bug that not all fields of EMC_TRAJ_STAT have cms->update() function executed ?
[11:40:29] <jepler> micges: sorry, in that case I misread it and I approve of that part of the change
[11:41:13] <jepler> can you configure your mailreader to send patches as text/plain instead of application/octet stream? this makes it easier to view the patch in my mailreader, application/octet-stream requires extra steps to view the contents.
[11:41:55] <micges> yes sure
[11:43:55] <CIA-48> EMC: 03jepler 07TRUNK * 10emc2/src/emc/task/emccanon.cc:
[11:43:55] <CIA-48> EMC: patch to new G64 Pn Qn mode
[11:43:55] <CIA-48> EMC: apparently the wrong tolerance value was being used. (patch from micges)
[11:45:29] <BigJohnT> jepler are you going to allow pushes for emc2-experimental?
[11:47:46] <jepler> BigJohnT: there will be something that allows pushes before we fully switch, to give developers a sense of how that works and to shake out the bugs
[11:48:01] <jepler> whether that's emc-experimental or just a sandbox which doesn't start with any useful contents, I'm not sure
[11:48:22] <BigJohnT> ok, either one so we can practice a bit
[11:51:37] <jepler> at this point it looks like it will be possible to run git entirely on dreamhost.com servers, which should lead to faster performance for everybody with more than 384kbps download speed. but that means more setup and more stuff for us to learn, compared to the system hosted in somebody's basement like we've done for the past few years
[11:52:10] <micges> jepler: did you know something about my later questions?
[11:53:19] <jepler> yes it's probably a bug not to have update() calls, but we have zero experts in cms/nml for you to get a sure answer from
[11:53:44] <micges> ah ok
[11:56:14] <BigJohnT> have you done "git-add -i"?
[11:56:36] <BigJohnT> thats cool
[12:00:30] <jepler> it's a bit cumbersome, but nice if you need it
[12:01:27] <BigJohnT> I'm just kicking the tires and pushing all the buttons to see if I get what I expect or I get some experiance :)
[12:02:53] <jepler> does qgit or git-gui have a way to help you do what 'git add -i' does? I can't even find the commit button in qgit..
[12:03:38] <BigJohnT> I have not found a way to add a file in qgit or git-gui yet
[12:04:04] <BigJohnT> it is in the Edit menu
[12:07:14] <jepler> in "git gui", when you select a new file from "unstaged changes", you can add it by selecting "Commit > Stage to commit"
[12:07:36] <jepler> you can do that to add all changes in an existing file, or you can right click in the patch view and "stage hunk for commit"
[12:08:54] <BigJohnT> cool
[12:09:55] <BigJohnT> you can click on a file in "Staged Changes" and it unstages it
[12:10:24] <jepler> lots of new terms to learn like "staged"
[12:13:52] <BigJohnT> * BigJohnT is off to work
[12:13:56] <BigJohnT> talk to you later
[12:14:47] <jepler> I should too
[12:34:36] <skunkworks_> so - is there only going to be backlash after the conversion? ;)
[12:44:30] <jepler> Why would there be?
[12:44:35] <jepler> everyone has a chance to discuss it now
[12:45:07] <jepler> it seems to me that only a person who is more interested in stirring up controversy would deliberately wait until after a decision is final to discuss it
[12:45:45] <skunkworks_> heh
[15:37:01] <skunkworks_> who's on first?
[15:46:25] <cradek> wtf?
[17:13:22] <micges> alex_joni: hello
[17:13:52] <micges> do you have any machine controlled by emc2 ?
[17:36:04] <BJT-Work> what's on second
[18:22:21] <jepler> Subject: [Rtai] Is RTAI-Lab able to manage 3 exits?
[18:22:25] <jepler> I wonder what that even means
[19:33:59] <micges> jepler: I have question: in what dir do you want to go with run from line functionality ?
[19:43:13] <micges> my opinion: I have few oxygen cutting machines, plasma cutter is building, few lasers are running emc2 from about year and they all require RFL that will continue running selected vector regards of current position
[19:46:50] <cradek> can you describe more exactly what behavior is needed?
[19:47:32] <micges> I'll try
[19:50:46] <micges> when you stop machine, position where was abort is saved, I can jog to see how cutting is going, then when you run, machine will goto to that saved position, then resuming executing that same gcode that was stopped
[19:51:27] <cradek> go to the saved position in a straight line?
[19:51:44] <micges> we talking of XY machine
[19:52:05] <micges> yes straight line
[19:52:27] <cradek> if the solution is for an XY only machine, it cannot be incorporated into the main emc2
[19:52:59] <micges> I understand
[19:53:00] <cradek> needs to make sense for mill/lathe too
[19:53:30] <cradek> do you think there is a more general solution that is good for mill/lathe/xy?
[19:53:52] <micges> I have solution for mill but can't imagine of lathe
[19:54:02] <micges> didn't have any
[19:54:46] <micges> for mill for 2,5d and 3d safe trip is working correctly
[19:55:08] <micges> that means goto max z, goto saved xy, then goto saved z
[19:55:50] <cradek> I think that is usually ok for a mill, but not always
[19:56:00] <cradek> lathe is harder
[19:57:23] <cradek> even for XY only, I think it is hard
[19:57:36] <cradek> things you can change while the machine is off will disrupt the path
[19:57:54] <cradek> if you are in an arc, you need to start from the beginning of the arc, or generate a new arc
[19:58:19] <cradek> if you change tool diameter, the path moves and the saved location is not even on it
[19:58:20] <micges> cradek: I've already programed XY solution ;)
[19:58:28] <micges> it works, even for arcs
[19:59:20] <cradek> that is impressive then
[19:59:36] <cradek> can you change the tool while it is stopped?
[20:00:06] <micges> didn't tried
[20:00:24] <micges> now I'm trying understand naivecam
[20:00:50] <micges> it doesn't operate with my code
[20:01:39] <micges> there are some assumtions of naivecam code that I don't know if they are intentionally or not
[20:02:47] <jepler> you also have to consider o-words, probing, inputs/outputs, the state of interpreter parameters
[20:03:04] <jepler> even if the result of considering them is "define them as non-working or undefined with rfl"
[20:03:25] <jepler> but consider that the *current* rfl is the result of someone writing what satisfied his needs of the moment, and now it doesn't fulfill yours
[20:03:26] <cradek> I think that would have to be some of them, yes
[20:04:06] <cradek> it was much easier to get right back when our gcode was simple and linear (not that we had it right, even then)
[20:04:07] <jepler> also, we've rehashed this a billion times, I wish there was a wiki page so we wouldn't have the same discussion once a month until the sun burns out
[20:04:33] <cradek> jepler: new ideas are always good - I'm interested in fixing this too - I just don't know how.
[20:04:52] <cradek> until I know how it should work, I don't want to touch it :-/
[20:04:53] <cradek> brb
[20:04:59] <micges> jepler: I don't have this discussion that far
[20:09:49] <micges> I'll check things you mention to consider with rfl
[20:16:19] <micges> jepler: cradek: thanks for helping
[20:16:22] <micges> bbl
[20:36:03] <micges> good night