> > Some of us developers felt that breaking out many of these > components into smaller > pieces that communicated through some sort of switchboard > that was easily adjusted > would be a big improvement. Some developers thought this > would overly complicate > EMC and cause performance issues. This was back when EMC > was still run on Pentium > Classic computers. > > John Kasunich set to work and produced a very elegant code > which did add some > complexity, but had almost ZERO performance impact -- all > data was passed > with pointers, so no extra overhead when running. And, it > did make it MUCH easier > to add components to do special stuff. Most of the > developers thought it was great, > and a solution to many issues that were just too difficult > to deal with. >
I apologize for being repetitive, but everytime a discussion on this list touches this topic I can't feel anything else but admiration for what you the developers did and keep doing with LinuxCNC. But specially with things like this that break huge barriers. For me as a mortal (at least in the programming world), seing a thing like HAL working is pure art and I consider it the pinnacle of pragmatism. I still need to make the jump and fully submerge myself in the world of programming, at least to better understand more of the code that make LCNC work. Someday maybe.. Again sorry for being repetitive, but the majority of the things seen here I think are from another level. I guess the open source community in general could be like this, but I speak from my experience here. As always thank you! El lun., 30 mar. 2020 a las 22:45, Jon Elson (<el...@pico-systems.com>) escribió: > On 03/30/2020 07:17 PM, John Dammeyer wrote: > >> On 03/30/2020 04:27 PM, John Dammeyer wrote: > >>> When did the hal come in? This stuff has dates on or around > 2004--2007. > >>> > >>> > >> John Kasunich proposed HAL at a meeting in 2005, I'm pretty > >> sure, and demonstrated a > >> small version of it in 2006, I think. It was merged into > >> EMC some time after that, probably > >> late 2006 to 2007. > >> > >> Jon > >> > > Thanks Jon, > > For some reason I thought it was a lot earlier. > > > A major roadblock had developed in adding new things to EMC > about 2002 or so. > There was no way to have axes slave to the spindle, for > instance. Since all of EMC > was just a few massive chunks of code, making any but the > most trivial change in > the real time side was a HUGE hassle. I know -- for my > boards, i needed to move > the auxiliary I/O (spindle, coolant etc.) over to the real > time side, as everything communicating > with my boards needed to be coordinated through a single > thread. It took mods to > about a dozen different files to get it done. > > Some of us developers felt that breaking out many of these > components into smaller > pieces that communicated through some sort of switchboard > that was easily adjusted > would be a big improvement. Some developers thought this > would overly complicate > EMC and cause performance issues. This was back when EMC > was still run on Pentium > Classic computers. > > John Kasunich set to work and produced a very elegant code > which did add some > complexity, but had almost ZERO performance impact -- all > data was passed > with pointers, so no extra overhead when running. And, it > did make it MUCH easier > to add components to do special stuff. Most of the > developers thought it was great, > and a solution to many issues that were just too difficult > to deal with. > > The logjam was broken, and a whole bunch of new features and > capabilities were > realized. > > Jon > > > _______________________________________________ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > _______________________________________________ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
