Am 20.10.2012 um 22:18 schrieb Jon Elson: > Jon Elson wrote: >> Michael Haberler wrote: >> >>> the - somewhat brittle - run-from-line feature has seen several fixes >>> related to this, both by cradek and me, in 2.5, see >>> >>> >>> >> I have another machine with a brand-new build from master, so I will >> test on that one. >> I should have thought of doing that first! >> > Brittle is right! I tested it on the system I just built from master > last week, 2.6.0~pre, > and it does exactly the same thing. If I select the line with the G01 move, > and do run from line, it performs that move, but never executes the > subroutine. > It sits there in some in-between state, forever. > > If anyone wants to work on it, the snippet of code I posted on 10/16 > is all you need. The subroutine waits a second, does a small -Z move, > turns on "mist coolant" for 2 seconds, waits 2 seconds, turns off > coolant, waits a second and does a +Z move and returns. If you > want a longer copy of the program, email or reply to this message, > I can send it.
Re 'working on it': no doubt someone could, somehow, patch the interpreter into submission to handle this or that case; I dont think this improves the overall situation. the fundamental issue is that the concept of source location is still in it's pre-subroutine, pre-multiple gcode files state, and without a fundamental rework of "line numbers" into a more realistic "source location" which refers to file, linenumber, file type, and call context (definition or reference) this is not going to bring about better quality. This is a quite far-reaching change because it goes all the way from NGC source file to motion and back - all this is currently handled through an integer line number which eventually mutates into a 'motion id'. I've sketched some ideas about source location on the wiki a while ago which outline the principle; it needs updating though. It's one of the two major prerequisites I'd like to see adressed before carrying over any interpreter into LinuxCNC3; the other one being generalized handling of variable references (see QueuebustersRevisited on the wikie). - Michael > > I'm afraid this behavior probably goes pretty deep into interpreter > code that I have no understanding of, so I wouldn't have much luck > researching it myself. > > Thanks, > > Jon > > ------------------------------------------------------------------------------ > Everyone hates slow websites. So do we. > Make your web apps faster with AppDynamics > Download AppDynamics Lite for free today: > http://p.sf.net/sfu/appdyn_sfd2d_oct > _______________________________________________ > Emc-users mailing list > [email protected] > https://lists.sourceforge.net/lists/listinfo/emc-users ------------------------------------------------------------------------------ Everyone hates slow websites. So do we. Make your web apps faster with AppDynamics Download AppDynamics Lite for free today: http://p.sf.net/sfu/appdyn_sfd2d_oct _______________________________________________ Emc-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/emc-users
