#emc-devel | Logs for 2011-03-15

Back
[00:01:28] -!- theorbtwo has quit [Read error: Operation timed out]
[00:08:18] -!- acemi has quit [Quit: WeeChat 0.3.2]
[00:19:06] -!- The_Ball has quit [Ping timeout: 250 seconds]
[00:37:40] -!- adb has quit [Ping timeout: 252 seconds]
[00:45:35] <andypugh> seb_kuzminsky: I think I see the problem with that SPI wiki page.
[00:47:02] <andypugh> You can't register the tram and load the channel descriptors at setup time, as the pins which set them don't have any values until they are netted and the feeder-functions have run. And that all happens far too late in the sequence.
[00:48:19] -!- adb [[email protected]] has joined #emc-devel
[00:54:32] -!- skunkworks [skunkworks!~chatzilla@str-bb-cable-south2-static-6-412.dsl.airstreamcomm.net] has joined #emc-devel
[01:02:55] -!- andypugh has quit [Quit: andypugh]
[01:03:03] -!- PCW has quit [Quit: ChatZilla 0.9.86 [Firefox 3.6.13/20101203075014]]
[01:07:07] -!- crazy_imp has quit [Ping timeout: 248 seconds]
[01:16:13] -!- adb has quit [Ping timeout: 246 seconds]
[01:16:35] -!- adb [[email protected]] has joined #emc-devel
[01:55:41] -!- tlab has quit [Ping timeout: 255 seconds]
[02:18:27] -!- tlab has quit [Quit: Leaving]
[02:19:15] <jepler> you can never decide what pins or parameters to create based on values seen on pins and parameters. You can only create pins and parameters until you're ready (component calls hal_ready), and halcmd loadrt blocks until the component is ready. If it didn't block, then at least sometimes the subsequent command in the hal file, such as a setp or a net, could execute before the pin or parameter was created (this was a cause of a lot
[02:19:45] <cradek> of a lot ...
[02:21:01] <jepler> (this was a cause of a lot of transient startup errors on very old emc)
[02:28:53] -!- KimK has quit [Quit: Leaving]
[03:07:05] -!- Ze1982 has quit [Quit: Leaving]
[03:18:03] -!- adb has quit [Ping timeout: 240 seconds]
[03:31:25] -!- cblack001 has quit [Remote host closed the connection]
[03:33:55] -!- cblack001 [[email protected]] has joined #emc-devel
[04:02:20] -!- KimK [[email protected]] has joined #emc-devel
[05:13:51] -!- Athlocatle has quit [Ping timeout: 246 seconds]
[05:38:08] davec_ is now known as Guest66940
[05:40:30] -!- emc2-buildmaster has quit [Ping timeout: 240 seconds]
[05:40:54] -!- cevad has quit [Ping timeout: 240 seconds]
[05:41:01] -!- seb_kuzminsky has quit [Ping timeout: 260 seconds]
[05:50:27] -!- seb_kuzminsky [[email protected]] has joined #emc-devel
[06:07:07] -!- psha [[email protected]] has joined #emc-devel
[06:22:27] -!- ve7it has quit [Remote host closed the connection]
[06:23:06] -!- kb8wmc has quit [Quit: ChatZilla 0.9.86 [Firefox 3.6.15/20110303170030]]
[07:03:03] -!- jstenback has quit [Quit: ZNC - http://znc.sourceforge.net]
[07:03:29] -!- bdale has quit [Ping timeout: 255 seconds]
[07:09:58] <psha> cradek: i believe i've found way to avoid mass rebuild
[07:10:03] -!- pjm__ has quit [Ping timeout: 252 seconds]
[07:16:21] -!- logger[mah] has quit [Ping timeout: 246 seconds]
[07:16:21] -!- steves_logging has quit [Ping timeout: 246 seconds]
[07:17:03] -!- cv has quit [Ping timeout: 246 seconds]
[07:18:27] -!- bdale has quit [Ping timeout: 246 seconds]
[07:19:44] -!- logger[mah] [logger[mah][email protected]] has joined #emc-devel
[07:26:20] -!- pingufan has quit [Remote host closed the connection]
[07:30:12] <psha> cradek: http://psha.org.ru/cgit/psha/emc2.git/commit/?h=doc-fixes
[07:30:36] <psha> http://psha.org.ru/cgit/psha/emc2.git/commit/?h=doc-fixes&id=41094d
[07:30:43] <psha> this is better
[07:37:40] -!- roberth__ [[email protected]] has joined #emc-devel
[07:56:56] -!- psha has quit [Read error: Operation timed out]
[07:57:19] -!- psha [[email protected]] has joined #emc-devel
[08:40:59] -!- cradek has quit [Ping timeout: 248 seconds]
[08:49:07] -!- cradek [[email protected]] has joined #emc-devel
[09:09:49] -!- tris has quit [Excess Flood]
[09:14:27] -!- pingufan has quit [Remote host closed the connection]
[09:19:12] -!- mrsun has quit [Remote host closed the connection]
[09:35:49] -!- nullie has quit [Read error: Connection timed out]
[09:39:10] -!- Techrat has quit [Quit: Leaving]
[09:46:46] -!- cv has quit [Ping timeout: 250 seconds]
[09:49:22] -!- mk0 has quit [Ping timeout: 240 seconds]
[09:53:15] -!- mozmck has quit [Read error: Operation timed out]
[09:54:54] -!- mk0_lan_died has quit [Quit: Leaving]
[09:57:43] -!- mozmck [mozmck!~moses@client-173.225.233.220.dfwtx.partnershipbroadband.com] has joined #emc-devel
[10:04:06] -!- acemi has quit [Changing host]
[10:04:44] -!- pingufan has quit [Remote host closed the connection]
[10:05:50] uwe__ is now known as uwe_
[10:11:26] -!- seb_kuzminsky has quit [Ping timeout: 240 seconds]
[10:13:46] -!- seb_kuzminsky [[email protected]] has joined #emc-devel
[10:30:30] -!- adb [[email protected]] has joined #emc-devel
[10:38:06] -!- adb has quit [Ping timeout: 240 seconds]
[10:46:08] -!- adb [[email protected]] has joined #emc-devel
[11:22:24] -!- pjm has quit [Ping timeout: 246 seconds]
[11:45:28] -!- cradek has quit [Ping timeout: 250 seconds]
[11:47:30] -!- cradek [[email protected]] has joined #emc-devel
[12:02:05] -!- skunkworks has quit [Ping timeout: 276 seconds]
[12:18:52] -!- cjdavis has quit [Ping timeout: 250 seconds]
[12:25:10] <psha> reposting for sure
[12:25:17] <psha> cradek: http://psha.org.ru/cgit/psha/emc2.git/commit/?h=doc-fixes&id=41094d
[12:25:31] <psha> cradek: possible fix for mass docs rebuild on every change
[12:27:46] -!- psha has quit [Quit: Lost terminal]
[12:35:39] -!- SteveStallings [[email protected]] has joined #emc-devel
[12:35:56] SteveStallings is now known as steves_logging
[12:37:31] -!- skunkworks [skunkworks!447329d2@gateway/web/freenode/ip.68.115.41.210] has joined #emc-devel
[12:39:49] -!- isssy has quit [Quit: Visitor from www.linuxcnc.org]
[12:50:09] -!- kb8wmc has quit [Quit: ChatZilla 0.9.86 [Firefox 3.6.15/20110303170030]]
[12:59:00] <cradek> psha: thanks, I will check it out
[13:08:26] -!- pingufan has quit [Remote host closed the connection]
[13:08:26] -!- cjdavis1 has quit [Ping timeout: 255 seconds]
[13:18:11] -!- mk0 has quit [Quit: Leaving]
[13:28:33] -!- i_tarzan has quit [Read error: Operation timed out]
[13:51:21] <cradek> psha: when I touch touchy.txt, I still get a big rebuild: asciidoc processes axis.txt (and others), axis's images are converted again, etc
[14:04:13] -!- cblack001 has quit [Remote host closed the connection]
[14:10:42] -!- psha[work] [psha[work][email protected]] has joined #emc-devel
[14:12:04] <psha[work]> cradek: hm, need to check
[14:47:51] -!- Connor has quit [Ping timeout: 246 seconds]
[14:59:18] <psha[work]> cradek: hm, i've just checked and everything is ok
[14:59:26] <psha[work]> touch ../docs/src/gui/touchy.txt
[14:59:35] <psha[work]> got rebuild for Master_User.pdf and touchy.html
[15:03:41] -!- Athlocatle has quit [Quit: �����]
[15:10:11] -!- pcw_home has quit [Remote host closed the connection]
[15:13:28] <psha[work]> bb
[15:13:45] -!- cjdavis has quit [Ping timeout: 252 seconds]
[15:13:54] -!- psha[work] has quit [Quit: leaving]
[15:14:05] -!- skunkworks has quit [Ping timeout: 252 seconds]
[15:20:18] -!- skunkworks [skunkworks!447329d2@gateway/web/freenode/ip.68.115.41.210] has joined #emc-devel
[15:27:17] cjdavis1 is now known as cjdavis
[15:36:26] -!- psha [[email protected]] has joined #emc-devel
[15:40:19] <cradek> "Opens up the HAL Configuration widow"
[15:41:43] -!- VisualEcho has quit [Quit: http://quassel-irc.org - Chat comfortably. Anywhere.]
[16:12:25] -!- ve7it [[email protected]] has joined #emc-devel
[16:17:45] -!- awallin [[email protected]] has joined #emc-devel
[16:21:55] -!- JT-Work [[email protected]] has joined #emc-devel
[16:22:30] -!- awallin has quit [Ping timeout: 252 seconds]
[16:23:26] -!- awallin [[email protected]] has joined #emc-devel
[16:30:12] -!- awallin has quit [Ping timeout: 252 seconds]
[16:34:53] -!- awallin [[email protected]] has joined #emc-devel
[16:47:51] <CIA-5> EMC: 03cradek 07v2.5_branch * r7076f06a5cbf 10/docs/src/gui/axis.txt: markup fixes
[16:47:53] -!- psha has quit [Quit: -a-]
[16:47:53] <CIA-5> EMC: 03cradek 07v2.5_branch * r05b26cbc93dc 10/docs/src/gui/axis.txt: docs: update and clean AXIS docs
[16:47:59] <CIA-5> EMC: 03cradek 07v2.5_branch * r41094df1ffee 10/docs/src/Submakefile: docs: Prevent mass rebuilds on any change
[17:03:19] -!- VisualEcho has quit [Quit: http://quassel-irc.org - Chat comfortably. Anywhere.]
[17:35:27] -!- pcw_home has quit [Remote host closed the connection]
[18:04:27] <jepler> is anyone planning to work on fixing hardy package building for v2.5? If not, we should ask seb to drop the package-building variants from the buildbot.
[18:08:20] <seb_kuzminsky> i think psha mentioned some fix...
[18:08:35] <jepler> I'm the first to admit I haven't been paying much attention
[18:09:50] <cradek> master-package-rt-hardy-source looks ok...?
[18:10:01] <cradek> where should I be looking?
[18:10:02] <seb_kuzminsky> it can make the dsc, but not the deb
[18:10:15] <seb_kuzminsky> psha said it had to do with the location of configs, and "new style tables"
[18:10:28] <jepler> enode/ip.68.115.41.210] has quit [Ping timeout: 252
[18:10:28] <jepler> seconds]
[18:10:30] <jepler> argh
[18:10:33] <jepler> http://buildbot.linuxcnc.org/buildbot/builders/deb-hardy-sim-binary-i386/builds/572/steps/shell_4/logs/stdio
[18:10:53] <seb_kuzminsky> look here, then click on one of the red "build dep failed" stdio links: http://buildbot.linuxcnc.org/buildbot/waterfall
[18:10:58] <jepler> a2x: ERROR: /usr/bin/asciidoc --backend docbook -f ../docs/src/docbook.conf --doctype book --attribute scriptdir=../docs/src/ --verbose --out-file /tmp/buildd/emc2-2.5.0~pre1-65/docs/src/Master_HAL.xml /tmp/buildd/emc2-2.5.0~pre1-65/docs/src/Master_HAL.txt returned non-zero exit status 1
[18:11:07] <cradek> gotcha
[18:11:14] <jepler> possibly due to: ERROR: tutorial.txt: line 922: unsafe: {sys:../docs/src/image-wildcard "/tmp/buildd/emc2-2.5.0~pre1-65/docs/src/images/halscope-11.png" /tmp/buildd/emc2-2.5.0~pre1-65/docs/src }
[18:12:31] <jepler> possibly fixed by --unsafe or --asciidoc-opts=--unsafe
[18:13:18] <jepler> but that's asciidoc version dependent. 8.5.0 "Added --safe option and deprecated --unsafe option" (making the former --unsafe the default if I understand what it's saying)
[18:13:48] <jepler> (which would be why lucid with 8.5.2 works and hard with 8.2.2 fails)
[18:14:27] -!- psha [[email protected]] has joined #emc-devel
[18:15:24] <jepler> psha: seb said maybe you had a fix for the hardy build errors?
[18:15:57] <psha> yea, i've just checked logs
[18:16:30] <psha> besides safe/unsafe options you've mentioned i think it's mainly caused by 'new style tables'
[18:16:53] <psha> one option is to place 'asciidoc' script just like we've did with a2x
[18:17:12] * jepler makes a sound like he's been punched in the gut
[18:17:25] <jepler> I thought we'd picked a tool with a stable input format
[18:17:36] <jepler> but two versions two years apart aren't even compatible
[18:17:57] <psha> i've mentioned that when we've discussed that earlier - new style tables are really useful :(
[18:18:25] * psha feels shamed
[18:18:31] <jepler> well, personally I am not bothered at all by not supporting 8.04 for emc 2.5
[18:18:49] <psha> now our bottom line is 8.5 i guess
[18:19:18] <psha> however asciidoc document way to make 'standalone' distribution - prepare configs and place ./asciidoc somewhere around
[18:19:44] <psha> so you get something not depending on installed/not installed version
[18:21:14] <jepler> hopefully this situation is different than what we'll face at 12.04
[18:21:52] <jepler> in this case, we wrote and tested on systems with newish asciidoc (10.04 and squeeze) and then found that we'd used features that don't exist or don't work as we require them by default in the two-years-older version of asciidoc
[18:22:07] <psha> not even 8.5, but 8.3
[18:22:25] <jepler> that doesn't mean we should predict that 12.04 will have a version of asciidoc that won't process these files just fine
[18:22:29] -!- nullie has quit [Quit: Ex-Chat]
[18:22:56] <psha> i believe that it's strictly forward compatible
[18:22:58] <jepler> I *really* am uneasy at copying more parts of the doc building toolchain into our source tree; we'll get stuck on an outdated version and face more pain later
[18:23:23] <psha> yes, that's only workaround - not a true solution
[18:23:59] <jepler> so even after the immediate problem (unsafe execution of scripts) is fixed, asciidoc on ubuntu 8.04 will encounter other errors or format the docs improperly?
[18:25:10] -!- mhaberler [[email protected]] has joined #emc-devel
[18:25:10] <psha> hm, i've seen errors regarding 'header' attribute of tables
[18:25:55] <psha> FAILED: /tmp/buildd/emc2-2.5.0~pre1-65/docs/src/hal/comp.txt: line 112: illegal [paradef-default] options: header: header
[18:26:27] <jepler> can you remind me why we decided to include a copy of a2x?
[18:27:03] <psha> encoding issues on pre-8.6 versions
[18:27:22] <psha> due to bug in htmllib and missing workaround in a2x
[18:27:42] <jepler> right, thanks.
[18:30:10] <cradek> I still see recommendations that people try u8 when u10 doesn't install or run properly (too many CPUs, too old, apic problems)
[18:30:23] <cradek> our u8 kernel is the last one that seems to run on everything
[18:30:37] <jepler> so when we start building on wheezy or natty, we'll want to add a configure check for working a2x/htmllib so that we can use the platform version in preference to the (potentially older) bundled version
[18:31:07] <psha> cradek: newer kernels are better on newer hw but 8.04 is usually reported as only working option
[18:31:11] <jepler> cradek: you're the 2.5 release manager, so if you call for us to make sure 2.5 packages are buildable on hardy I think we should listen.
[18:32:12] <psha> btw is it possible to use 8.04 kernel on 10.04?
[18:32:39] <cradek> doubt the packaging is compatible because of the grub changes
[18:33:02] <cradek> jepler: I guess it comes down to how much pain is involved in keeping it working
[18:33:03] <psha> hm, grub is not related to kernel - it's just bootloader
[18:33:17] <cradek> psha: I know, but the deb twiddles the grub config
[18:33:19] <psha> cradek: not much - either place local asciidoc/a2x scripts
[18:33:40] <psha> or (for package building) backport 8.6 asciidoc package
[18:33:59] <psha> i usually choose second option for my work
[18:36:51] <psha> cradek: btw i've checked mass-rebuild-fix and it seem to work
[18:37:05] <psha> on clean copy
[18:37:24] <cradek> I will try again then
[18:37:25] <cradek> thanks
[18:39:46] * jepler is checking whether the lucid asciidoc will build on hardy
[18:39:51] -!- x_ [[email protected]] has joined #emc-devel
[18:39:51] -!- JT-Work has quit [Quit: ChatZilla 0.9.86 [Firefox 3.6.15/20110303024726]]
[18:40:03] <jepler> .. it does
[18:40:08] x_ is now known as micges
[18:42:51] <jepler> now to see if that will build pdf docs in master..
[18:47:51] <jepler> asciidoc: Master_HAL.txt: line 7: evaluating: {sys:../docs/src/image-wildcard "/usr/src/emc2/docs/src/common/emc2-intro.*" /usr/src/emc2/docs/src }
[18:47:54] <jepler> Traceback (most recent call last):
[18:47:57] <jepler> File "../docs/src/image-wildcard", line 38, in <module>
[18:47:59] <jepler> print os.path.relpath(i, docpath)
[18:48:02] <jepler> AttributeError: 'module' object has no attribute 'relpath'
[18:48:15] <jepler> os.path.relpath is new in version 2.6
[18:48:44] <cradek> ffs
[18:48:51] -!- andypugh [andypugh!~andy2@cpc2-basl1-0-0-cust1037.basl.cable.virginmedia.com] has joined #emc-devel
[18:49:32] -!- x_ [[email protected]] has joined #emc-devel
[18:49:39] -!- micges has quit [Ping timeout: 252 seconds]
[18:49:45] x_ is now known as micges
[18:49:47] <jepler> psha: as an aside, do you know whether it's possible to make this an error instead of a warning as far as asciidoc is concerned?
[18:49:54] <jepler> asciidoc: WARNING: Master_HAL.txt: line 7: {sys:../docs/src/image-wildcard "/usr/src/emc2/docs/src/common/emc2-intro.*" /usr/src/emc2/docs/src }: non-zero exit status
[18:51:22] -!- mk0 has quit [Quit: Leaving]
[18:53:53] <psha> hm, need to check
[18:54:17] <psha> jepler: re: relpath, i've fixed version but have not yet commited it
[18:54:23] <jepler> I have a fix here as well
[18:54:30] <jepler> copy implementation from posixpath.py
[18:54:44] <psha> since i'm using same script on translation wiki
[18:55:45] <jepler> so we think that asciidoc 8.5 is the minimum that can build docs?
[19:01:17] <jepler> here's what I have so far:
[19:01:19] <jepler> http://emergent.unpy.net/files/sandbox/0001-docs-provide-implementation-of-relpath.patch
[19:01:22] <jepler> http://emergent.unpy.net/files/sandbox/0002-docs-check-for-supported-asciidoc-version.patch
[19:04:26] <jepler> :( still build errors
[19:04:27] <jepler> pdflatex failed
[19:04:27] <jepler> Master_User_tmp.tex:214: Missing $ inserted.
[19:04:27] <jepler> Master_User_tmp.tex:214: leading text: ...Typical_machine}{"[fig:Typical_machine]"}
[19:04:59] -!- rizo has quit [Read error: Operation timed out]
[19:08:59] <andypugh> Can I poke people for some opinions on my SPI question?
[19:09:53] <cradek> unfortunately my only opinion is that you know much more about it than I do
[19:10:11] <jepler> fixing that by using a different character than underscore, I get another error
[19:10:19] <andypugh> Which solution looks neatest?
[19:10:26] <jepler> Master_Integrator_tmp.tex:1574: Missing $ inserted.
[19:10:26] <jepler> Master_Integrator_tmp.tex:1574: leading text: [amplifier[volts]=(output[\frac{mm}{sec}
[19:11:12] <cradek> just change it to ascii mm/sec?
[19:11:38] <jepler> there are some more equations in that file, would have to change them all
[19:12:17] <jepler> we'll want to change them anyway, though -- there's no effort to lay them out as equations in the html docs :( http://linuxcnc.org/docs/devel/html/config/ini_config.html
[19:12:34] <jepler> andypugh: I am so out of touch that I don't know what the spi proposals are
[19:13:07] <andypugh> I posted a long and rambling post to the dev list today.
[19:13:25] <jepler> [Msgs:12918 New:1020]
[19:13:30] <jepler> I'm a little behind :-P
[19:14:06] -!- redwil has quit [Remote host closed the connection]
[19:14:49] <jepler> andypugh: this is clearly insane: bspi_chans=0x1000055F.0x90010057
[19:15:16] <jepler> andypugh: and this can't work, because the ability of modules to read arbitrary files in userspace is limited and for all practical purposes should be considered nonexistent: bspi_conf=7i65.0.ini
[19:15:22] <andypugh> Really? I was coming round to the idea.
[19:15:57] <andypugh> So, do we go for insane or impossible?
[19:16:25] -!- seb_kuzminsky has quit [Quit: Client exiting]
[19:16:27] <jepler> what are those numbers in bspi_chans? can you tell me how a user would figure them out?
[19:16:43] -!- seb_kuzminsky [[email protected]] has joined #emc-devel
[19:17:17] <andypugh> They are a bitfield which defines the bit length, echo, chip-select lines etc.
[19:17:30] <jepler> I think the fundamental mistake of the "spi sub-driver" page is that it is trying to do everything through HAL
[19:18:08] <andypugh> Definined in page 14 of http://www.mesanet.com/pdf/motion/7i65man.pdf
[19:19:03] <andypugh> My idea was that when the 7i65 (or any other SPI HAL module) started up it would spit it's required config into dmesg.
[19:19:22] <jepler> the communication between the core hostmot2 driver and an spi driver for a specific SPI daughtercard should be through C API calls
[19:19:33] <jepler> that's the model I think is sane
[19:19:54] <jepler> isn't the design of SPI that each distinct card has a distinct identification value that can be read?
[19:20:14] <seb_kuzminsky> jepler: no :-(
[19:20:20] <andypugh> But the sequence of events is just completely wrong for HAL communications to work.
[19:20:29] <seb_kuzminsky> the mesa "smart serial" protocol has that nice feature, but spi does not
[19:20:42] <seb_kuzminsky> andypugh: i agree with jeff, hal is not the way to go
[19:20:47] <seb_kuzminsky> the wiki page is way off in the weeds
[19:23:47] <andypugh> Hmm, so, the suggestion is that there should be a function embedded in the Hostmot2 SPI driver code which is called by the comp-based sub-driver during its setup phase which defines the channel configs and registers the TRAM.
[19:25:04] <jepler> here's a total draft of the way a C API would work: http://pastebin.com/QG5HcYRD
[19:25:16] -!- EDocTooR [[email protected]] has joined #emc-devel
[19:26:22] <jepler> so when the main hostmot2 finds that a particular spi port should be configured for a bogoboard, it calls bogoboard_setup, which can create pins and parameters
[19:26:44] <jepler> it calls the other functions at the appropriate times as well, for instance before the general tram write and after the general tram read
[19:28:37] <seb_kuzminsky> jepler: i had imagined it a bit backwards from that
[19:28:53] <andypugh> That requires that the Hostmot2 driver is aware of all SPI devices. It probably makes more sense for the Hostmot2 driver to simply create placeholders when it finds it has SPI functions in the bitfile, and have all the board "knowledge" in the board-specific hal module
[19:29:06] <seb_kuzminsky> i'd imagined the hostmot2 driver would make available a function like hm2_register_spi_subdriver()
[19:29:11] <jepler> andypugh: no, it doesn't. hm2_bugboard is a separate hal component
[19:29:24] <jepler> er, bogoboard in the example
[19:29:35] <seb_kuzminsky> and bogoboard.ko would call it, passing in a struct of bogoboard functions & spi setup info
[19:29:43] <jepler> seb_kuzminsky: isn't that what I wrote?
[19:29:48] <seb_kuzminsky> is it?
[19:29:50] * seb_kuzminsky rereads
[19:30:02] <seb_kuzminsky> ok sure
[19:30:04] <cradek> if seb and jepler agree, I'm with them, whatever the hell they're talking about
[19:30:18] -!- maximilian_h [[email protected]] has joined #emc-devel
[19:30:18] -!- maximilian_h has quit [Client Quit]
[19:30:33] <seb_kuzminsky> you said "when the main hostmot2 finds that a particular spi port should be configured for a bogoboard ..."
[19:30:48] <seb_kuzminsky> if that means, "bogoboard registers itself with the main hostmot2" then i think we agree
[19:30:57] <jepler> seb_kuzminsky: that's referring to 'config=...spi=bogoboard.bogoboard.cleverboard'
[19:31:01] <andypugh> Right, so the main Hostmot2 driver "finds that a SPI port should be configured for bogoboard" when bogoboard.ko says "Oi, HM2, over here, make me some pins and make is snappy!"
[19:31:28] <jepler> the main hostmot2 would look through its list of registered spi modules until it found a name that matched.
[19:31:34] <seb_kuzminsky> huh
[19:31:43] <seb_kuzminsky> yeah that's opposite than what i thought
[19:32:10] <seb_kuzminsky> i thought hostmot2 would notice its spi ports (config=...,num_spi=$NUM_SPI)
[19:32:11] <andypugh> I think putting anything specific in the modparam parsing is probably a problem, unless all it does is make a placeholder and wait for a matching card name to come along and talk to it.
[19:32:48] <andypugh> The aim is that anyone can add a new SPI device without touching the Hostmot2 code.
[19:32:53] <jepler> and it should be noted that my sketch has no way to pass parameters to "the bogoboard on spi channel 1 of card 2 connector 3"
[19:33:13] <seb_kuzminsky> and then you'd "loadrt bogoboard spi=hm2_5i20.0.spi.4"
[19:33:34] <jepler> seb_kuzminsky: ah, hmmm
[19:33:52] <seb_kuzminsky> or make a new kind of hal object called a spi_endpoint or something
[19:34:13] <seb_kuzminsky> and hook the hm2_5i20.0.spi.4.spi_endpoint to bogoboard.0.spi_endpoint
[19:34:22] -!- mhaberler has quit [Quit: mhaberler]
[19:34:24] <seb_kuzminsky> with a pointer in HAL
[19:34:32] <seb_kuzminsky> but maybe i'm smoking crack on that one
[19:34:55] <andypugh> I think you are if you think I am capable of coding that :-)
[19:35:07] <jepler> I think that's enough harder than what we've been discussing that we shouldn't do it right now
[19:35:22] <seb_kuzminsky> fair enough
[19:36:23] <andypugh> I think I see a way forward, and as a definite plus it saves us wiring several dozen HAL pins between the HM2 module and the HAL function.
[19:36:40] <jepler> yes, it'll shift a *lot* of burden away from the guy writing the hal files
[19:36:57] <jepler> that guy already has it rough as it is
[19:38:19] <skunkworks> they are usually a bunch of whiners
[19:42:21] <jepler> seb_kuzminsky: as long as you think parsing the spi port designation isn't too hard (spi=hm2_5i20.0.spi.4) I think yours is the better idea because it doesn't have the problem of passing options to the spi sub-driver
[19:42:55] <jepler> either way it seems like there'll be a structure of function pointers that will tell the core hostmot2 what to call for each spi
[19:44:12] <jepler> or is there a way to turn that inside out too and I just don't see it yet?
[19:44:48] <seb_kuzminsky> you know me, i love parsing text in the kernel
[19:44:56] <jepler> because you're kraaaaazeee
[19:45:18] <seb_kuzminsky> yeah i'm "loco", "dude"
[19:45:57] <seb_kuzminsky> so hm2_spi_register() would take a couple of arguments
[19:46:03] <jepler> bbl, time for a cup of coffee
[19:46:07] <seb_kuzminsky> one is the anyio board and spi endpoint to connect to
[19:46:21] <seb_kuzminsky> another is a struct of function pointers in the spi-device subdriver to call as needed
[19:46:29] <seb_kuzminsky> i think that'd work? maybe?
[19:46:32] <seb_kuzminsky> let us know, andypugh !
[19:46:37] <seb_kuzminsky> coffee, good idea
[19:46:39] <seb_kuzminsky> bbl
[19:46:54] <jepler> andypugh: have a cup of coffee with us
[19:50:34] <jepler> so the basic parts to test develop/test an spi sub-driver would be 7i46 + 7i50? is it feasible to use a microcontroller like an AVR as an SPI slave instead of adding unneeded boards?
[19:51:15] <jepler> or is it possible to jumper for SPI loopback?
[20:02:45] <andypugh> Loopback is probably an option.
[20:04:02] <andypugh> I am not sure that function pointers are needed, I was simply going to pass a pointer to a structure to the sub-driver, in just the same way that Hostmot2 passes the hm2 structure to all the sub-functions.
[20:05:09] <andypugh> In fact, all that structure needs to be is an array of 16 u32s.
[20:05:39] <andypugh> (I think) It is stall taking shape in my head
[20:15:44] <jepler> hm .. the toolchain running on 10.04 to produce pdfs does manage to format those equations as equations
[20:15:48] <jepler> but when producing html it doesn't
[20:15:56] <jepler> and when running on 8.04 it also doesn't
[20:29:26] -!- adb has quit [Ping timeout: 240 seconds]
[20:31:32] -!- mozmck has quit [Ping timeout: 250 seconds]
[20:32:01] -!- x_ [[email protected]] has joined #emc-devel
[20:32:53] -!- adb [[email protected]] has joined #emc-devel
[20:33:16] <jepler> it looks like some formulas process properly but others don't
[20:33:18] -!- micges has quit [Ping timeout: 250 seconds]
[20:33:35] <jepler> I wonder if it's dependent on the version of dblatex?
[20:36:57] -!- adb has quit [Read error: Operation timed out]
[20:38:22] <jepler> block math chockes, inline math is fine
[20:39:00] -!- adb [[email protected]] has joined #emc-devel
[20:42:07] <jepler> hmmmmm
[20:42:08] <jepler> >> http://dblatex.sourceforge.net/doc/release-notes/sec-changes-0210.html
[20:42:20] <jepler> release 0.2.10 of dblatex (hardy has 0.2.6, lucid has 0.2.12)
[20:47:19] -!- mozmck [mozmck!~moses@client-173.225.233.219.dfwtx.partnershipbroadband.com] has joined #emc-devel
[20:54:49] <psha> jepler: re: fail on failed subcommands
[20:55:07] <psha> it's not possible to treat failed subcommands as errors
[20:56:03] <psha> http://code.google.com/p/asciidoc/source/browse/asciidoc.py#795[3~
[20:56:07] <psha> http://code.google.com/p/asciidoc/source/browse/asciidoc.py#795
[20:56:36] x_ is now known as micges
[20:57:55] <jepler> I would love to know their justification for that practice
[21:00:07] <psha> i believe that they accept patch for something like '--strict' option
[21:01:13] <jepler> > Figure 3-3 shows the flow of the example application. The application is implemented by polled function calls to make it less dependable of other modules.
[21:01:21] <jepler> </application note>
[21:02:28] <skunkworks> jepler: welcome back :)
[21:07:06] <psha> jepler: it's possible to get math in html
[21:07:28] <psha> not as pretty as in pdf though
[21:08:01] <jepler> [qanda] also seems to be broken in the dblatex of ubuntu 8.04
[21:08:22] <jepler> it generates this LaTeX which is bad because you can't have a paragraph end inside \textit:
[21:08:25] <jepler> \begin{enumerate}
[21:08:28] <jepler> \item{}\textit{
[21:08:30] <jepler> I get an unexpected unmap event in my handler function right after startup. What's this?
[21:08:34] <jepler> }
[21:08:36] <jepler> (there's an extra blank line before } that isn't shown in the paste I just made)
[21:09:06] <jepler> while on 10.04 you get \item{}{\it...}
[21:09:07] <psha> pity but not so bad
[21:09:36] <jepler> || pdflatex failed
[21:09:37] <jepler> /usr/src/emc2/src/Master_Integrator_tmp.tex|4173| Paragraph ended before \text@command was complete.
[21:09:42] <psha> is it dblatex or asciidoc?
[21:09:42] <jepler> it's a fatal error so I think it's pretty bad
[21:09:45] <jepler> dblatex
[21:09:58] <jepler> I'm about to look into also building 10.04's dblatex on 8.04..
[21:10:01] <psha> i mean not so bad == [qanda] is not very valuable
[21:11:00] <psha> it was transfered from wiki but really that's not something to live in docs
[21:12:01] <jepler> huh, this is the first time I've seen a debian source package with multiple orig.tar files
[21:12:06] <jepler> http://archive.ubuntu.com/ubuntu/pool/main/d/dblatex/dblatex_0.2.12-4.dsc
[21:12:24] <jepler> .. won't build on 8.04: dpkg-source: error: Unsupported format of .dsc file (3.0 (quilt))
[21:12:28] <psha> wow
[21:12:38] <psha> really strange :)
[21:14:38] <jepler> well, as long as we're committed to just excising every construct that offends dblatex 0.2.6 I think we'll be able to get something built on 8.04
[21:15:29] <jepler> the asciidoc package built from the lucid sources seems to work, it's dblatex problems that I'm hitting now
[21:15:53] <psha> asciidoc package is really trivial - bunch of configs and two scripts
[21:16:04] <jepler> though I think I know where to find the bit that generates \textit .. it's in a dblatex xsl file
[21:16:08] <jepler> we could fix it and repackage
[21:16:11] <jepler> rather than upgrading
[21:16:42] <jepler> or in fact we could try the .12 orig.tar.gz with the .6 dsc and patch too
[21:19:37] <jepler> wow the git on ubuntu 8.04 is old
[21:19:53] <psha> patching old version is more simple then repackaging new
[21:25:04] -!- mhaberler [[email protected]] has joined #emc-devel
[21:29:24] <CIA-5> EMC: 03jepler 07master * r138a4b335a98 10/docs/src/drivers/pico_ppmc.txt: fixup lyx conversion
[21:29:25] <CIA-5> EMC: 03jepler 07master * re966f3cc68af 10/docs/src/config/emc2hal.txt: document new tool offsets, and give hints for g95
[21:29:26] <CIA-5> EMC: 03jepler 07master * ra0795d04f022 10/docs/src/config/ (ini_config.txt ini_config_fr.txt): docs: Explicitely terminate lists in config/ini_config
[21:29:27] <CIA-5> EMC: 03jepler 07master * r345a2efd249a 10/docs/src/config/ini_homing.txt: docs: new parameters that affect homing
[21:29:28] <CIA-5> EMC: 03jepler 07master * r306f6c49d0af 10/docs/html/gcode.html: docs: try again to get G10 right in the quickref
[21:29:29] <CIA-5> EMC: 03jepler 07master * r41094df1ffee 10/docs/src/Submakefile: docs: Prevent mass rebuilds on any change
[21:29:35] <CIA-5> EMC: 03jepler 07master * r05b26cbc93dc 10/docs/src/gui/axis.txt: docs: update and clean AXIS docs
[21:29:47] <CIA-5> EMC: 03jepler 07master * r7076f06a5cbf 10/docs/src/gui/axis.txt: markup fixes
[21:29:47] <CIA-5> EMC: 03jepler 07master * r87d59902c686 10/src/configure.in: docs: check for supported asciidoc version
[21:29:48] <CIA-5> EMC: 03jepler 07master * r97bf8ca70f4d 10/debian/control.in: call for minimum asciidoc version in build-deps
[21:29:49] <CIA-5> EMC: 03jepler 07master * r5eeeae0fbdda 10/docs/src/common/user_intro.txt: docs: underscore in reference is bad
[21:29:49] <CIA-5> EMC: 03jepler 07master * rd02639d62043 10/ (12 files in 8 dirs): Merge branch 'v2.5_branch'
[21:30:10] <jepler> OK, this still doesn't get docs building on 8.04 but it gets us closer. still need to get my asciidoc package public, and either get rid of the remaining constructs that are bad for old dblatex or also get a patched dblatex available
[21:30:17] <jepler> bbl
[21:30:55] <andypugh> Does anyone fancy pushing my changes to the smart-serial driver, or the gearchange component?
[21:36:25] -!- cjdavis has quit [Read error: Connection reset by peer]
[21:37:58] <psha> jepler: cool
[21:45:12] <psha> andypugh: you've mentioned that bit set/show widget would be useful
[21:45:33] <psha> but how to set up it's layout? horiz/vert? grouping?
[21:45:56] <andypugh> Horizontal, I would thing, LSB to the right?
[21:50:47] <psha> so it would not be useful as addition in axis
[21:51:57] <andypugh> Hmm, it would fit in a GladeVCP, but not in the PyVCP pane arranged that way
[21:52:37] <andypugh> I have rather forgotten why I wanted it now.
[21:52:39] -!- skunkworks has quit [Ping timeout: 252 seconds]
[21:56:30] <andypugh> Every time that the main Hostmot2 driver calls a sub-function it passes it a pointer to a hosmot_t structure. Is it allowable for a sub-function to hold on to that pointer in a static variable so that it can access the data structure when other functions call it, in turn? (or, put another way, can the structure be pulled from a static rather than pushed)
[21:58:32] -!- cjdavis has quit [Ping timeout: 255 seconds]
[21:58:39] -!- e3m has quit [Quit: OUCH!!!]
[21:59:17] -!- x_ [[email protected]] has joined #emc-devel
[21:59:18] <psha> andypugh: it may be done as hook for some container widget
[21:59:32] <psha> i don't think it would be widely useful
[21:59:58] <psha> i'll show you example tomorrow
[22:00:47] <andypugh> I am going to see if it works, but my question is "is it bad"
[22:02:02] -!- micges has quit [Ping timeout: 276 seconds]
[22:03:38] <mhaberler> can I request 2 trivial merges: 1) fixing iocontrol man page - add missing pin 2) add missing file for toolchange-in-gcode demo 'emc owordm6-ui.ini'
[22:03:40] <mhaberler> http://git.mah.priv.at/gitweb/emc2-dev.git/shortlog/refs/heads/fix-iocontrol-manpage
[22:03:56] <mhaberler> and http://git.mah.priv.at/gitweb/emc2-dev.git/shortlog/refs/heads/add-missing-osubdemo-ui-file
[22:04:45] -!- tlab has quit [Quit: Leaving]
[22:13:54] <psha> bb all
[22:14:29] <andypugh> Night
[22:14:39] <mhaberler> cu
[22:15:24] x_ is now known as micges
[22:16:10] -!- ewidance [[email protected]] has joined #emc-devel
[22:16:18] -!- psha has quit [Quit: Lost terminal]
[22:19:20] -!- e3m [[email protected]] has joined #emc-devel
[22:22:58] <andypugh> I am about to delete a bunch of code and start down a different path. Is there a neat way in Git to keep this stuff in case the new path is a dead end?
[22:23:42] <micges> make a new branch based on this one
[22:23:59] <andypugh> OK, then checkout the old one?
[22:25:09] <micges> you will be still in old one
[22:26:42] <micges> but you will have two identical branches at this point
[22:26:57] <micges> so one is archive one is new devel
[22:27:11] <JT-Shop> git branch my-topic master
[22:27:32] <JT-Shop> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Git#Work_with_branches
[22:27:53] <andypugh> I doubt I will want it back, so I will assume I can if necessary :-)
[22:28:18] <JT-Shop> always nice to save things for a rainy day
[22:30:36] -!- x_ [[email protected]] has joined #emc-devel
[22:33:06] -!- micges has quit [Ping timeout: 260 seconds]
[22:33:06] -!- mhaberler has quit [Ping timeout: 240 seconds]
[22:33:12] x_ is now known as micges
[22:33:48] <seb_kuzminsky> andypugh: the hostmot2_t tracks the state of a particular instance of the hostmot2 firmware, ie of a particular anyio board
[22:34:15] <seb_kuzminsky> you can only use that particular hostmot2_t if you're doing something on behalf of that particular anyio board
[22:34:30] <andypugh> Ah.
[22:35:05] <micges> andypugh: do you use all drivers you've wrote?
[22:35:16] <andypugh> micges: None of them :-)
[22:35:59] <micges> hehe
[22:36:05] <andypugh> I probably will be using the 8i20 stuff, they are good cards and I seem to have two of them which cost me nothing :-)
[22:39:17] <andypugh> seb_kuzminsky: I have decided not to hold the hm2 anyway, but I am holding a copy of the individual bspi module. I guess that needs to be an array of bspi modules.
[22:42:06] <andypugh> Is anybody going to get upset if I make that a fixed-size array of them? It only needs to persist between the hm2 driver loading, and the sub-driver asking it to make pins.
[22:43:56] <andypugh> I wonder how many FPGA boards is feasible?
[22:45:22] <andypugh> seb_kuzminsky: Have you made any assumptions about the maximum board count?
[22:45:55] <seb_kuzminsky> i think not in hostmot2.c, but probably in hm2_pci.c and hm2_7i43.c
[22:47:39] <andypugh> Yeah, 8 PCI and 4 P-Port
[22:49:09] <andypugh> You realise that means you can never have more than 16896 IO pins until Pete brings out his SSerial splitter?
[22:53:30] -!- Fox_Muldr has quit [Ping timeout: 276 seconds]
[23:01:59] -!- uwe_ has quit [Ping timeout: 255 seconds]
[23:12:23] -!- e3m_ [[email protected]] has joined #emc-devel
[23:14:30] -!- e3m has quit [Ping timeout: 250 seconds]
[23:18:41] -!- skunkworks [skunkworks!~chatzilla@str-bb-cable-south2-static-6-412.dsl.airstreamcomm.net] has joined #emc-devel
[23:30:47] -!- mikegg has quit [Ping timeout: 255 seconds]
[23:37:33] -!- ewidance has quit [Quit: ewidance]
[23:38:04] -!- e3m_ has quit [Quit: Clap on! , Clap off! Clap@#&$NO CARRIER]
[23:39:12] -!- seb_kuzminsky has quit [Ping timeout: 252 seconds]
[23:41:38] -!- micges has quit [Quit: Ex-Chat]
[23:50:53] -!- seb_kuzminsky [[email protected]] has joined #emc-devel