#linuxcnc-devel Logs

Jan 02 2020

01:15 PM seb_kuzminsky: i'm about to release 2.7.15
01:18 PM seb_kuzminsky: it's been a year and a half since 2.7.14
02:00 PM JT-Shop: time flies when your having fun
04:53 PM jepler: good afternoon
04:54 PM jepler: pcw_home: pcw_mesa: I forget, did you get a chance to check the pin naming issue that micges said existed on https://github.com/LinuxCNC/linuxcnc/pull/665 ?
04:54 PM jepler: I'm tempted to go ahead and merge that into master and pick up any pieces that break off.
04:56 PM jepler: still trying to figure out the best kernel packaging workflow, the build itself works based on JT's instructions and can be completed in way under 1 hour on a PI4 with HDD (and which doesn't store temporary files on SD!) so .. I think I should just make a debian package that does that. But starting Debian packaging from scratch is never^Walways fun
04:56 PM jepler: (some *names* did get fixed in the second batch of updates, but I didn't do anything about numbering)
04:57 PM jepler: aannd.. I broke the CI. :checks:
05:04 PM jepler: oh looks like some stuff changed upstream too and it interacts badly
05:06 PM pcw_mesa: The 7I95 pin names look OK to me (at least my current code)
05:07 PM jepler: okay
05:07 PM jepler: I'm not sure which date I last took a copy of newmesa.zip from you. It was the one that corrected some pin name strings, I think
05:08 PM jepler: f.ex fixing the name of BISS pins
05:08 PM jepler: https://github.com/LinuxCNC/linuxcnc/pull/665/commits/7fff17be27f95e8d7344613f2d6aa5a07227157d
05:08 PM pcw_mesa: Yeah that was the last thing
05:08 PM jepler: OK
05:09 PM jepler: I also made some changes where values of type "float" are used in hostmot2 https://github.com/LinuxCNC/linuxcnc/pull/665/commits/67ea53e21be502991d91035f9a5473e4e6549ead which you might want to incorporate back if we can't get the "newmesa.zip" workflow stamped out again :)
05:09 PM jepler: (this is while we were thinking there might be a precision problem of absolute encoders, but it's not addressing any known problem)
05:11 PM pcw_mesa: Ahh well I'm hoping such a "forklift upgrade" is never done again...
05:14 PM pcw_mesa: though maybe I will get trustworthy enough (to not break stuff) and get push access I don't know C but I can copy/paste from my betters
05:18 PM pcw_mesa: I'm working on expanding SSerial to 7 registers (224 bits) for SP6 and above only as it requires SRL32's instead of SRL16s for the UART FIFOs
05:22 PM pcw_mesa: Had to double the CPU program address range ( by making a simple instruction set optimization )
05:23 PM jepler: pcw_home: as far as I'm concerned, you can have commit access if you want it
05:24 PM jepler: I understand worrying about the subtleties of modern C, but mostly I deal with that by ... not worrying about it too much :) and anyway, you know how to test the hostmot2 hal driver
05:26 PM pcw_mesa: Ill have to get used to git (though I have used it with JTs mesaflash fork)
05:27 PM pcw_mesa: well not really using git other than pulls commits and pushes
05:30 PM jepler: Yeah, I think using git can be the bigger hurdle
05:32 PM jepler: hey wait what is this. "make deb-pkg" !?
05:33 PM jepler: (I bet it doesn't yield a usable package, but don't disillusion me yet!)
05:40 PM jepler: skunkworks: the content of the file should give the CPU speed of the RT CPU core: /sys/devices/system/cpu/cpufreq/policy0/scaling_cur_freq
05:41 PM jepler: assuming we are still scheduling RT tasks on the highest numbered CPU core
05:41 PM jepler: should not be hard to make a userspace python component that outputs this value on a pin, if CPU frequency is still in question
05:51 PM cerna: jepler: "I understand worrying about the subtleties of modern C" -> Can you give an example? (I have a feeling that C is one of the easiest languages.)
05:53 PM jepler: cerna: https://blog.regehr.org/archives/213 this blog is one of my regular reads, and this is just one of many articles about C being surprising
05:54 PM jepler: Let’s look at another Type 2 function:
05:54 PM jepler: int stupid (int a) {
05:54 PM jepler: return (a+1) > a;
05:54 PM jepler: }
05:54 PM jepler: .. the compiler is free to convert this into a function that always returns 'true', despite a programmer's intention
05:55 PM jepler: (you might have believed that INT_MAX + 1 is a negative number and therefore less than a; but modern standards state that this interpretation is wrong)
05:57 PM cerna: Aha, OK. Thanks. We have different definition of "quirks". (I learned to not presume anything the hard way, so I am "factory" trained.)
07:15 PM skunkworks: jepler: thanks - I will take a look