Back
[09:40:33] <mozmck> seb_kuzminsky: It would be really nice if the website search also searched IRC logs
[10:14:36] <Tom_itx> is there an 'official' log site?
[10:18:29] <archivist> many unofficial methinks
[10:18:38] <Tom_itx> yes
[10:22:26] <seb_kuzminsky> mozmck: yeah that would be great
[10:23:10] <mozmck> yes, I guess as far as that goes, it would also be nice if IRC logs were on linuxcnc.org too.
[10:25:03] <seb_kuzminsky> yeah, how that should probably work is this
[10:25:16] <seb_kuzminsky> somebody commits to running their logger on wlo
[10:25:27] <seb_kuzminsky> somebody transfers existing logs in a compatible format to wlo
[10:25:49] <seb_kuzminsky> ... err, done (at that point google will index it since its on wlo, and the search will find it)
[10:25:50] <jepler> I think running irc bots is forbidden by dreamhost policy
[10:25:57] <seb_kuzminsky> haha
[10:25:59] <archivist> merge logs to fix gaps
[10:26:55] * seb_kuzminsky whistles and walks away
[10:30:48] <mozmck> Hmm, maybe the bot can run somewhere else and just store the log on wlo?
[10:31:01] <mozmck> I don't know how the bots work really.
[10:31:26] <archivist> mine is a php bot storing in mysql
[10:33:07] <archivist> some store in text files
[10:33:41] <mozmck> I think toms and psha's use text files - which are easily searchable.
[10:34:08] <mozmck> jepler: you are correct:
https://www.dreamhost.com/legal/acceptable-use-policy/
[10:36:50] <Tom_itx> you're welcome to my logs if you can use them
[10:37:40] <Tom_itx> it's working at least _most_ of the time
[10:39:03] <Tom_itx> http://tom-itx.no-ip.biz:81/~tom-itx/irc/logs/%23linuxcnc/index.html
[10:39:06] <Tom_itx> back to 2012
[10:39:37] <Tom_itx> http://tom-itx.no-ip.biz:81/~tom-itx/irc/logs/%23linuxcnc-devel/index.html
[10:39:42] <Tom_itx> and mid 2013 for devel
[10:40:17] <Tom_itx> i doubt anything before that is relevant now anyway
[10:41:46] <jepler> only for very long games of "I told you so"
[10:42:35] <mozmck> I don't know - I'm browsing through some of the old NIST emails finding some information on how things work in linuxcnc :-)
[10:43:20] <archivist> I have #emc going back to 2010
[10:44:24] <Tom_itx> mine start at 2011 for that one
[10:45:11] <Tom_itx> * ChanServ sets mode: +l 1
[10:45:14] <Tom_itx> final entry
[10:45:47] <archivist> rare styff like #emc-devel | seb_kuzminsky | 2010-08-10 21:06:06 | i can not ssh to it, best i have is a vnc to its locked-up console
[10:48:14] <archivist> it would be "fun" merging data across timezones with all the offsets and gaps and delays on the irc servers
[10:58:04] <jepler> jthornton: I ****built from source**** on linuxmint-17.2-xfce-64bit.iso and AXIS works fine. As I said before, I believe your mistake is using a package for a different linux distribution.
[10:59:03] <jepler> (all inside a qemu virtual machine)
[11:03:48] <mozmck> I have not had any problem with axis either, but I'm not using rtai either. I also built from source.
[11:07:46] <jepler> this was uspace against a vanilla kernel, but rtai is probably not the problem
[11:08:04] <jepler> at one point jt stated he was using a package from some other OS and I believe that's the problem
[11:11:40] <pcw_home> Yeah I built linuxcnc with Mint 17.2 and it all seemed to work ok (uspace also)
[12:39:40] <Roguish> pcw_home: hey, how did you do the Mint OS? a script or ??? I have tried and Axis does not work. JT has the same problem.
[12:41:35] <mozmck> Roguish: are you installing the precise packages?
[12:41:58] <Roguish> all: there seems to be an anomaly in the .ini parser. in the [Traj] section if I add a '#' and comment after the parameter on a line the parser reads the # and comment as more parameters. do the same in [Axis] sections and every thing is cool.....
[12:42:41] <Roguish> mozmck: the pricise packages??? I have followed the description by JT on the forum.
[12:43:23] <Roguish> https://forum.linuxcnc.org/forum/18-computer/29917-linux-mint-with-rtai?start=10#65548
[12:43:27] <mozmck> I think that installs packages built for ubuntu precise. pcw_home, jepler, and I have all built linuxcnc from source and it works fine.
[12:43:36] <jepler> the steps I did to build and install linuxcnc on linuxmint-17.2-xfce-64bit.iso:
http://paste.debian.net/334690/
[12:43:44] <jepler> uspace and I did not install any realtime kernel, but AXIS works
[12:44:03] <jepler> INSTALLING A PACKAGE BUILT FOR A DIFFERENT DISTRO WILL PROBABLY HAVE STUFF BROKEN AND THAT IS NOT A BUG.
[12:44:52] <Roguish> mozmck and jepler: thanks. I will follow down thru your procedure. and check it out.
[12:45:56] <Roguish> jepler. in my previous line about the parser, that is in Master on a straight forward Debian.
[12:46:29] <Roguish> was / is a completely different topic from the Mint issues.
[12:48:15] <jepler> FOO = bar # this is not a comment
[12:48:31] <jepler> our inifile format is underdocumented, but it only supports whole-line comments
[12:48:47] <mozmck> I thought I read that somewhere...
[12:49:55] <jepler> it may be the case that some places that expect numbers ignore anything after an initial part that parses as a number(?)
[12:51:33] <jepler> for instance, making this change, linuxcnc still starts just fine
[12:51:34] <jepler> -BACKLASH = 0.000
[12:51:34] <jepler> +BACKLASH = 0.000ZZZZZZZZZZZZZZZZ
[12:53:43] <jepler> and with 1.0asdfsadfasdfasdf it acts just like 1.0 was specified
[12:55:45] <Roguish> jepler: i've been using in line comments for quite a while in most all sections.
[12:56:46] <Roguish> without issues. I was just consolidating some lines with comments and ran into the cited issue.
[12:59:53] <jepler> consider this hypothetical inifile entry: REGISTER_FOR_FEED_RATE = #1000
[13:00:26] <jepler> and you'll see that we cannot actually make # a comment character
[13:01:24] <jepler> above I explained that it is a quirk in number parsing in some parts of linuxcnc that probably gave you the impression that a # could be used in this context to denote a comment
[13:01:29] <jepler> but it was a false impression
[13:02:29] <Roguish> jepler; that's cool. a brief word in the docs might help. i definitely see why it cannot be done. as i stated, 'anomaly' not bug.... i will avoid it where it does not work. THANKS. I always appreciate the help.
[13:03:48] <jepler> ok, you're welcome.
[13:24:11] <jepler> this patch tightens up the inifile parser so that where a number is expected, trailing junk after the number is no longer ignored:
https://emergent.unpythonic.net/files/sandbox/0001-inifile-detect-spurious-trailing-text-on-numbers-in-.patch
[13:24:42] <jepler> with my bad BACKLASH= line as above, I get this at startup (and linuxcnc doesn't work): INIFILE: ERR_CONVERSION, section=AXIS_0, tag=BACKLASH, num=1, lineNo=143
[13:25:02] <jepler> I don't plan to push it to any branch, it's just for demonstration purposes
[13:35:35] <jthornton> jepler: thanks I think I understand now, I'll build a RIP later today
[13:45:59] <pcw_home> Roguish: as mozmck said I build linuxcnc from source on Mint and that works fine
[13:46:57] <pcw_home> ( I also built a Preemt-RT kernel and ran linuxcnc with real hardware on Mint )
[13:47:08] <Roguish> pcw_home: ok, i'm loading a new Mint now and will go thru jepler's install.
[13:47:11] <mozmck> Is there a way to tell task to synch interp from motion?
[13:48:00] <mozmck> It looks like I should probably NOT call emcTaskPlanSynch() from control.c
[13:48:29] <Roguish> jepler: i like that patch. it even tells the line number!!! you ought to push it. it's a good one.
[13:55:26] <jthornton> is there a configure switch to install?
[13:56:01] <mozmck> jthornton: no, you really should build packages to install
[13:56:03] <jepler> jthornton: you should either RIP or you should build a package. you shouldn't "make && make install".
[13:56:34] <jthornton> ok thanks
[15:02:41] <jthornton> ok dokey I finally got it and understand more about it now, thanks for your patience with me
[16:06:02] * JT-Shop just build a deb and installed it, not linuxcnc but a tiny step
[16:11:18] <jepler> to build for a rtai kernel that you've already booted into, it should suffice to run: debian/configure -r ; debuild
[16:11:28] <jepler> instead of the debian/configure uspace that I used
[16:30:13] <cradek_> seb_kuzminsky: could you remind me what you didn't like about the g52 branch?
[16:35:06] <KGB-linuxcnc> 03John Morris 05cradek/zultron/g52 a0cab80 06linuxcnc 10src/emc/rs274ngc/interp_convert.cc interp_convert.cc: noop: wrap ridiculously long lines * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=a0cab80
[16:35:06] <KGB-linuxcnc> 03John Morris 05cradek/zultron/g52 c77a255 06linuxcnc 10src/emc/rs274ngc/interp_convert.cc interp_convert.cc: use readable symbols for `switch(g_code)` * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=c77a255
[16:35:06] <KGB-linuxcnc> 03John Morris 05cradek/zultron/g52 376d014 06linuxcnc 10(24 files in 6 dirs) Implement G52 offsets * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=376d014
[16:35:08] <KGB-linuxcnc> 03John Morris 05cradek/zultron/g52 cfd1523 06linuxcnc 10(11 files in 3 dirs) fixup: incorporate Chris Radek's feedback * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=cfd1523
[16:35:12] <KGB-linuxcnc> 03Chris Radek 05cradek/zultron/g52 53937cb 06linuxcnc 10docs/html/gcode.html add G52 to the quickref * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=53937cb
[16:55:01] <Roguish> jepler: I am going thru your process for uspace on Mint. You use an amd.deb. I have an Intel 64. should I change or what???
[16:56:15] <jthornton> wow I did debian/configure -r and got successfully configured for 'LinuxMint-17.2'-'3.4-9-rtai-686-pae'..
[16:57:39] <Roguish> It seems like I should replace the 'amd64.deb' with 'i386.deb' yes??? no??? or what?
[16:57:49] <jthornton> opps missed the last par ; debuild
[16:58:41] <jthornton> oh wow it's building
[17:31:34] <Roguish> just ran into a 'failure' in the 'signing' portion.
[17:31:36] <Roguish> 'gpg: skipped "Sebastian..... ..... secret key not available'
[17:31:37] <Roguish> 'gpg: /temp/debsign/........................... clearsign failed: secret key not available'
[17:31:39] <Roguish> 'debsign: pgp error.....'
[17:31:40] <Roguish> 'build fatal error at line 1283'
[17:31:42] <Roguish> 'running debsign failed'
[17:33:32] <Roguish> ok, stuck. mr. wizard, help me.
[17:33:39] <Roguish> please
[17:38:57] <JT-Shop> what are you trying to do Roguish ?
[17:39:03] * JT-Shop heads inside
[17:39:12] <Roguish> still working on a Mint install.
[17:39:47] <Roguish> this one per jepler's process for uspace
[17:44:34] <mozmck> jthornton: to build debs you should be able to run: fakeroot debian/rules binary
[17:44:53] <mozmck> from your linuxcnc-dev directory - not the src directory
[17:45:15] <mozmck> You may have to install fakeroot and possibly something else - I don't know for sure.
[17:45:24] <jthornton> run fakeroot debian/rules first then run debian/configure -r ; debuild
[17:48:20] <Roguish> mozmck: is this signing thing required?
[17:50:20] <mozmck> jthornton: why are you running debian/rules first?
[17:50:51] <mozmck> I run debian/configure -r; then fakeroot debian/rules binary and when it's done you have packages.
[17:50:59] <jthornton> ok, thanks
[17:51:17] <jthornton> I really am guessing a lot lol
[17:51:51] <mozmck> Roguish: the signing should not be required
[17:52:23] <Roguish> ok. i suspected so.
[17:52:48] <mozmck> jthornton: debuild probably does the same thing as fakeroot debian/rules binary, but not completely sure.
[17:52:57] <Roguish> http://paste.debian.net/334690/
[17:53:05] <Roguish> that is jepler
[17:53:14] <Roguish> jepler's procedure.
[17:53:14] <jthornton> I got make: `debian/control' is up to date. when I ran fakeroot
[17:53:52] <mozmck> Roguish: I bet debuild is trying to sign the package after building. If you run "fakeroot debian/rules binary" it will not do that step.
[17:54:05] <jepler> if the .deb files were create in .. then you're fine
[17:54:14] <jepler> signing is not necessary just to install a package on your own system
[17:54:15] <mozmck> jthornton: "fakeroot debian/rules binary"
[17:54:31] <jepler> "debuild -uc -us" skips the signing steps
[17:54:40] <Roguish> jepler. thanks for coming back in. the debs were created.
[17:55:18] <jepler> amd64 vs i386 just means whether you installed a 64-bit OS or a 32-bit OS. It's not about the brand of your CPU.
[17:55:31] <Roguish> ok, added -uc -us
[17:56:01] <mozmck> jepler: I should probably use debuild - I'm signing my packages in a separate step right now.
[17:56:02] <Roguish> ah, ok. i will use the amd64 as the cpu is a 64bit.
[17:56:38] <mozmck> amd had 64 bit first, so it got called amd64 IIRC
[17:56:54] <Roguish> compiling again
[17:56:57] <jepler> debuild will automatically pick whatever the installed system is
[17:57:12] <jepler> you can of course install a 32-bit OS on a 64-bit system
[17:57:16] <jepler> the other way around doesn't work so hot
[17:57:55] <Roguish> where is the script doing this 'build' process? is it called 'build......'
[17:58:00] <Roguish> what directory.
[17:58:45] <mozmck> in the debian directory
[17:59:00] <mozmck> rules is the main script
[17:59:36] <Roguish> ok, i am just curious to scan it.
[18:04:33] <Roguish> probably have to add something into the sources list.
[18:04:55] <Roguish> to keep it updated (if it's possible with this type of build).
[18:07:52] <jthornton> holy cow it worked, I'm just stoked right now... I don't even care that I forgot the salt in the bread
[18:08:35] <Roguish> use salted butter, lots of it
[18:09:15] <mozmck> :-) I've often salted fresh buttered bread.
[18:09:18] <jthornton> naw, I'll just postpone the shrimp poboys till tomorrow and make more
[18:16:49] <jepler> my wife's family consistently makes their bread without salt and have for 40 years
[18:16:56] <jepler> luckily I was able to train her (my wife) out of it
[18:22:28] <jthornton> New Orleans style french bread would not be the same I assume without salt
[18:23:15] <jthornton> Leidenheimer French Bread is the best imho
[18:23:55] <jthornton> this is as close as I can come to Leidenheimer's
http://gnipsel.com/recipes/bread/french-bread.html
[18:24:43] <jepler> no, it's quite unsatisfactory if you're not accustomed to it
[18:26:47] <jthornton> an acquired taste I take it
[18:27:25] <jthornton> thanks again for helping me succeed in getting LinuxCNC installed in LinuxMint
[19:23:19] <jepler> ABC: always be compiling