Re: [Emc-users] Problem pncconf in master (patch attached)
On 10/30/2014 11:24 PM, Chris Morley wrote: Hi Carsten Thank you for the bug reports and the patch. Unfortunately i don't think the icon patch is quite right. Can you check if this one works? Hi Chris, I tried your patch it and it works for me, both RIP and .deb. Please push it to 2.7 (or feel free to wait for Carsten's sign-off if you like). I think Carsten just had the arguments to his diff in the wrong order, and ended up with a reversed patch I will look into the kernel check error next. I have some data for you on that. I see that pncconf uses the hal python module to learn about the system it's running on. This is a sensible thing to do. With linuxcnc compiled with --with-realtime=$RTAI (where RTAI is some appropriate RTAI path), running on an RTAI kernel, you get this: hal.is_kernelspace=1 hal.is_rt=1 hal.is_sim=0 hal.is_userspace=0 hal.kernel_version matches the uname With linuxcnc compiled with --with-realtime=uspace, running on an rt-preempt kernel: hal.is_kernelspace=0 hal.is_rt=1 hal.is_sim=0 hal.is_userspace=1 hal.kernel_version is not set With linuxcnc compiled with --with-realtime=uspace, running on a non-rt-preempt kernel: hal.is_kernelspace=0 hal.is_rt=0 hal.is_sim=1 hal.is_userspace=1 hal.kernel_version is not set So in other words, in all three of the expected situations, the hal module has good, accurate data. I did not test the unexpected situations: --with-realtime=rtai on a non-rtai kernel and --with-realtime=uspace on an rtai kernel. Also, I tested Carsten's pncconf python module install patch[0][1] and it makes pncconf from the debian packages work better (still needs something like your icon path patch). And it does not break RIP builds. I just pushed that patch to 2.7. 0: https://sourceforge.net/p/emc/bugs/400/ 1: http://psha.org.ru/irc/%23emc-devel/2014-10-30.html#16:29:23 -- Sebastian Kuzminsky -- ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Problem pncconf in master (patch attached)
On 10/30/2014 11:40 PM, Chris Morley wrote: Seb or Jeff There are two new checks in module HAL is_kernelspace is userspace Can you explain exactly what I can assume if each is true? is_rt and is_sim basically meant: can drive real hardware/ couldn't drive real hardware. I played around a bit and my conclusion is this: I think that if hal.is_rt is True, that implies that you can run hardware. This seems to hold true for both rtai builds and uspace builds. Looking at halmodule.cc:1152 and at the rtapi_is(3) manpage (both in 2.7) confirms this conclusion. -- Sebastian Kuzminsky -- ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Problem pncconf in master (patch attached)
I will look into the kernel check error next. I have some data for you on that. I see that pncconf uses the hal python module to learn about the system it's running on. This is a sensible thing to do. So in other words, in all three of the expected situations, the hal module has good, accurate data. I did not test the unexpected situations: --with-realtime=rtai on a non-rtai kernel and --with-realtime=uspace on an rtai kernel. Ok thanks for that info. It seems the real problem is the missing kernel version when using uspace. I'll dream up a patch for you to test with - but tomorrow - bed time :) Chris M -- ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Problem pncconf in master (patch attached)
On 10/29/14 8:47 AM, Carsten Presser wrote: Hi, When trying to run pncconf from the recent buildbot debian package it complains about missing python modules. Those modules are missing. They should be under /var/lib/python/pncconf. I guess the changes to pncconf are not yet reflected in the debian control files. I am no debian-package-expert, so I cant propose a patch for that. Hi Carsten, thanks for your bug report. I confirm that pncconf in 2.7.0~pre2 does not start up, instead it gives a error. I opened bug #400 to track this: https://sourceforge.net/p/emc/bugs/400/ Pncconf in 2.6.3 does start up in but produces warnings, i opened bug #399 to track that: https://sourceforge.net/p/emc/bugs/399 However, I also found a few critical typos in the pncconf python module. A patch for those is attached. Thanks for the patch, that's great! Chris Morley is the pncconf developer maintainer. Chris, does that patch look right to you? -- Sebastian Kuzminsky -- ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Problem pncconf in master (patch attached)
Hi, there is an third issue I had with pncconf. Someone else already had if bevore: https://github.com/machinekit/machinekit/issues/102 My suggested fix is to add another if-case @line 1584 in pncconf, in function 'check_for_rt' -- elif hal.is_userspace: return True -- With that change the test-axis button works again. I am not quite sure how the 'is_userspace'-bit is behaves on RTAI systems, so this might not be a complete fix. Kind Regards Carsten p.s. i am working on a patch for the debian packaging issue. Am 30.10.2014 16:28, schrieb Sebastian Kuzminsky: On 10/29/14 8:47 AM, Carsten Presser wrote: Hi, When trying to run pncconf from the recent buildbot debian package it complains about missing python modules. Those modules are missing. They should be under /var/lib/python/pncconf. I guess the changes to pncconf are not yet reflected in the debian control files. I am no debian-package-expert, so I cant propose a patch for that. Hi Carsten, thanks for your bug report. I confirm that pncconf in 2.7.0~pre2 does not start up, instead it gives a error. I opened bug #400 to track this: https://sourceforge.net/p/emc/bugs/400/ Pncconf in 2.6.3 does start up in but produces warnings, i opened bug #399 to track that: https://sourceforge.net/p/emc/bugs/399 However, I also found a few critical typos in the pncconf python module. A patch for those is attached. Thanks for the patch, that's great! Chris Morley is the pncconf developer maintainer. Chris, does that patch look right to you? -- Carsten Presser Geschäftsführer c...@rstenpresser.de Elektronikentwicklung und Vertrieb Frankfurter Strasse 1 66606 St.Wendel Telefon: 06851-1305585 Mobil: 0179-5594690 Internet: http://ca.rstenpresser.de EMail: c...@rstenpresser.de -- ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Problem pncconf in master (patch attached)
Date: Thu, 30 Oct 2014 17:11:36 +0100 From: c...@rstenpresser.de To: emc-users@lists.sourceforge.net Subject: Re: [Emc-users] Problem pncconf in master (patch attached) Hi, there is an third issue I had with pncconf. Someone else already had if bevore: https://github.com/machinekit/machinekit/issues/102 My suggested fix is to add another if-case @line 1584 in pncconf, in function 'check_for_rt' -- elif hal.is_userspace: return True -- With that change the test-axis button works again. I am not quite sure how the 'is_userspace'-bit is behaves on RTAI systems, so this might not be a complete fix. Kind Regards Carsten p.s. i am working on a patch for the debian packaging issue. Hi Carsten Thank you for the bug reports and the patch. Unfortunately i don't think the icon patch is quite right. Can you check if this one works? I will look into the kernel check error next. Chris M pncconf_bug1.patch Description: Binary data -- ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Problem pncconf in master (patch attached)
Date: Thu, 30 Oct 2014 17:11:36 +0100 From: c...@rstenpresser.de To: emc-users@lists.sourceforge.net Subject: Re: [Emc-users] Problem pncconf in master (patch attached) Hi, there is an third issue I had with pncconf. Someone else already had if bevore: https://github.com/machinekit/machinekit/issues/102 My suggested fix is to add another if-case @line 1584 in pncconf, in function 'check_for_rt' -- elif hal.is_userspace: return True -- Seb or Jeff There are two new checks in module HAL is_kernelspace is userspace Can you explain exactly what I can assume if each is true? is_rt and is_sim basically meant: can drive real hardware/ couldn't drive real hardware. Chris M -- ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
[Emc-users] Problem pncconf in master (patch attached)
Hi, When trying to run pncconf from the recent buildbot debian package it complains about missing python modules. Those modules are missing. They should be under /var/lib/python/pncconf. I guess the changes to pncconf are not yet reflected in the debian control files. I am no debian-package-expert, so I cant propose a patch for that. However, I also found a few critical typos in the pncconf python module. A patch for those is attached. Kind Regards Carsten -- Carsten Presser Geschäftsführer c...@rstenpresser.de Elektronikentwicklung und Vertrieb Frankfurter Strasse 1 66606 St.Wendel Telefon: 06851-1305585 Mobil: 0179-5594690 Internet: http://ca.rstenpresser.de EMail: c...@rstenpresser.de --- private_data.py 2014-10-29 15:41:16.361855079 +0100 +++ /home/awok/linuxcnc-dev/lib/python/pncconf/private_data.py 2014-10-29 01:41:14.270715001 +0100 @@ -60,13 +60,13 @@ if not os.path.isfile(self.WIZARD): self.WIZARD = os.path.join(/etc/linuxcnc/linuxcnc-wizard.gif) if not os.path.isfile(self.WIZARD): -self.WIZARD = os.path.join(/usr/share/linuxcnc/linuxcnc-wizard.gif) +self.LINUXCNCICON = os.path.join(/usr/share/linuxcnc/linuxcnc-wizard.gif) if not os.path.isfile(self.WIZARD): wizdir = os.path.join(os.path.abspath(BIN), ..) self.WIZARD = os.path.join(wizdir, linuxcnc-wizard.gif) self.ICONDIR = os.path.join(os.path.abspath(BIN), ..) -self.LINUXCNCICON = os.path.join(self.ICONDIR, linuxcncicon.png) +self.LINUXCNC = os.path.join(self.ICONDIR, linuxcncicon.png) if not os.path.isfile(self.LINUXCNCICON): self.LINUXCNCICON = os.path.join(/etc/linuxcnc/linuxcnc-wizard.gif) if not os.path.isfile(self.LINUXCNCICON): -- ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users