For example, the current version has a parameter that switches the direction of the second gear. It makes rather more sense to do that by simply having a negative ratio for reversed gears. To retain compatibility I need a rather untidy if (gear == 2 && reverse) speed = -speed.
EMC: 03jepler 07master * rca321ba7b7f3 10/src/emc/usr_intf/axis/scripts/hal_manualtoolchange.py: hal_manualtoolchange: withdraw popup if toolchange is aborted
EMC: 03jepler 07master * rccfa8c2037d6 10/tcl/bin/pickconfig.tcl: pickconfig: make files writable
EMC: 03jepler 07master * r2e7d4506a65e 10/src/ (Makefile Makefile.modinc.in): build: we want sane math
EMC: 03jepler 07master * r7459f5516ec6 10/src/rtapi/rtapi_math.h: rtapi_math.h: frexp, isinf, signbit
EMC: 03jepler 07master * rbe3a915d5b2e 10/src/rtapi/vsnprintf.h: vsnprintf: use ch() everywhere
EMC: 03jepler 07master * r9a4f76459386 10/src/rtapi/vsnprintf.h: vsnprintf: let others use small_digits, large_digits
EMC: 03jepler 07master * r85fc0e31b32f 10/src/rtapi/vsnprintf.h: vsnprintf: function to add a string to output
EMC: 03jepler 07master * r2ab3938c9b23 10/src/rtapi/vsnprintf.h: vsnprintf: print floats like %A, not F[...]
EMC: 03jepler 07master * r3fe8805143a8 10/scripts/decode_dmesg: decode_dmesg: decode hex floats in dmesg
EMC: 03jepler 07master * re2d7c7db71e5 10/src/emc/usr_intf/axis/scripts/hal_manualtoolchange.py: get rid of debugging print
hm when you estop during toolchange, something sane thing seems to happen (iocontrol deasserts "change" and all is well)
but when you abort during toolchange, everything freezes up until "changed" is asserted
imagine how the external ladder feels :)
(ie, I wonder how many people actually account for that scenario in the ladder logic)
or whatever they use
[13:14:17] <jepler> http://pastebin.ca/2006695
yeah, well, we could probably better define how either side of the toolchange can abort it, and what the state after the aborted toolchange is
but for hal_manualtoolchange, my concern of the moment, treating "changed went false" as "io wanted to abort the change" is just about enough
.. and it worked fine for estopping during toolchange
.. but not for aborting during toolchange
heh, excellent! :)
uh oh, buildbot is pissy
E: emc2-dev: debian-changelog-file-contains-invalid-email-address buildslave@lucid-rtai-i386
or maybe this is the real error
Running test: /home/farmer/buildbot/hardy-rtai-i386/hardy-i386-trunkish-realtime-rip/build/tests/abs.0
insmod: error inserting '/home/farmer/buildbot/hardy-rtai-i386/hardy-i386-trunkish-realtime-rip/build/rtlib/rtapi.ko': -1 Unknown symbol in module
hmm. how do you actually get to the build/test logs on the buildbot?
you can pick which confusing view you like, then try to find the related error somewhere in that view.
yeah, that's what I was doing, but I never got to an actual log (for this build)
[13:27:21] <jepler> http://emc2-buildbot.colorado.edu/buildbot/builders/hardy-i386-trunkish-realtime-rip/builds/71/steps/runtests/logs/stdio
looks like the problem will turn out to be WARNING: "isinf" [/home/farmer/buildbot/hardy-rtai-i386/hardy-i386-trunkish-realtime-rip/build/src/rtapi.ko] undefined!
alex_joni: sorry it took so long
ah, ok. the confusing waterfall view is - confusing
and the commit list message says the build failed, when actually it's not the build, but the tests, that failed
morficmobile: no problem
SWPadnos: the message comes from the trigger of the build, not from the builder itself
or something like that seb explained
sigh, I wish there was somebody who really cared about hardy (i.e., planned to run master on hardy) who I expected to have the expertise to fix this isinf problem
.. because I sure know I don't care about hardy
I have two of the three requirements
bbl for real this time
EMC: 03jepler 07master * r2be4c54d278e 10/src/rtapi/rtapi_math.h: rtapi_math: implement isinf for pre-gcc4.4 systems
EMC: 03jepler 07master * r20d714ead8dd 10/scripts/decode_dmesg: decode_dmesg: script should be marked executable
Chat8988 is now known as emcrules_mobile
jepler: wow, it's really cool to read manuals :)
[19:35:23] <jepler> http://www.ini.uzh.ch/~conradt/Projects/PencilBalancer/
jepler: that git "alias" is also done for lp I think
lp = launchpad
that's bzr, not sure how they do it
lp: is not only alias to launchpad but also some custom auth mechs
andypugh, (switching here for development talk) what do you mean by "put the driver in setup mode any time it is stopped "
what do you mean by stopped?
The sserial interface has two modes "running" and "stopped"
oh, and you have a HAL pin to control that
The actual driver always completes each thread, if that was worrying you.
yeah, I was wondering how you'd tell if it was stopped, when your driver functions weren't getting called :)
Which, with tens-of-milliseconds response times and handshaking makes for some fun programming.
You write a command to a register, then watch for the readback to clear. Then check the error flags, then do it all again. The response to (say) card-start can be 100mS.
ah, so you have a state machine in the configuration part of the driver
I have two (and potentially 40 total instances).
There is a per-port state machine that handles start and stop, and a per-card that handles parameter setting.
I am almost certainly going to move the per-card into the per-port and then have a "which card to program" pin for the whole port.
mhaberler: hardly :) bit you have questions - i'm here :)
mhaberler: only useless talks :) i'm limited to text-only communications :)
psha: re easy widgets acces: what I'm going to do is a) pass panel to hal_init(), b) add the other gtkwidgets in a second dict in makepins.GladePanel
what second dict?
psha: for the occasional gtk widget
you want to add gtkwidgets in hal init?!
i'll pull you branch and see
its in my brain, not in a branch
no... I want to be able to access all widgets easily for persistence in the user.py file
this is a basic stupidity test ;-)
then do what you want with provided glade/gtkbuilder objects
you again commited lot of junk
I want to parcel out the libglade/gtkbuilder issue into makepins
this is a work branch..
you've commited time.9 in man pages :)
well.. it's a manpage alright :)
mhaberler: i'm not talking about lib/python/touchy in some of previous branches :)
ok.. will read uo and behave
it's no harm but it's hard to switch barnches ;(
I think I am seeing why nobody lets me loose with commit access.
here is your reason: learning curve
that's why git's main worflow is pulling and not pushing :)
psha: anyway: I will go ahead as I think fit and let you throw out what you dont like
i don't like time.9 :)
seriously i've not yet inspected what's going on
have to stash current signal work first
i think i'll look on it tomorrow
or i have chances to loose signals work...
EMC: 03jepler 07master * raf40e5d29013 10/docs/man/man9/.gitignore: ignore generated file
jepler: while youngsters talk wise one listens :)