Am 24.04.2012 um 21:00 schrieb andy pugh: > On 24 April 2012 18:53, Michael Haberler <[email protected]> wrote: >> I decided to ignore the conventional wisdom that it cannot be done, and gave >> it a try. > > To be fair, it was only ever described as "hard" not "impossible" and
yes, but 'impossible' sounded infinitely better in the feature pissing contest;) > I think you have made some changes to the underlying structure? The basic idea is to snapshot a motion queue state on 'retract', including current position within a move, and switch to an alternate motion queue where you can do arbitrary moves, like the retract or recovery move. On 'recover', when you are where you left off, switch to the primary queue, and continue. Currently task doesnt even notice that something happened, it's all just a motion which tooks it tool a long time to complete. The whole thing assumes coordinated mode and remains in it. I think jogging can be sled in without much damage. For the bigger picture, one could take the 'keep this completely isolated within motion' or 'make task aware of it' views. Not sure yet. > Is changing the tool length possible when retracted. or does that > invalidate the original queue? This would be nice but cant be exclusively solved at the motion level right now; I just discussed this with Chris on linuxcnc-devel irc. The issue is that offsets are currently applied during interpreter runs in the canon layer; once they reach task and motion everything's said and done wrt offsets; changing it there to compensate against the earlier decision is bound to be a hit-and-miss game. However we agreed this is too early, and moving offsetting to the task/motion area would help. That means offsets would be applied in motion, and could be changed there. Once that is done, that should be possible. I dont think it is that hard, but some aspects I dont understand yet, like how rotation would fit in. It also affects how tool information is used (aka 'who sees what'): the interpreter starts out with some view of tool information, then proceeds; then motion might change some offsets; then interpreter mght continue as it was held up by queue full. It needs to be clearly thought through and spelled out. This was no issue so far. - Michael > -- > atp > The idea that there is no such thing as objective truth is, quite simply, > wrong. > > ------------------------------------------------------------------------------ > Live Security Virtual Conference > Exclusive live event will cover all the ways today's security and > threat landscape has changed and how IT managers can respond. Discussions > will include endpoint security, mobile security and the latest in malware > threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ > _______________________________________________ > Emc-users mailing list > [email protected] > https://lists.sourceforge.net/lists/listinfo/emc-users ------------------------------------------------------------------------------ Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ _______________________________________________ Emc-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/emc-users
