Python Interface makes race conditions - mayby
02 Jan 2024 13:28 #289557
by Aciera
Replied by Aciera on topic Python Interface makes race conditions - mayby
Cmorley:
zz912:I would also suggest not using HALUI MDI commands.
Ask the Gmoccapy devs to add INI MDI command ability to Gmoccapy.
Now I'm confused, why would Chris suggest NOT to use 'HALUI MDI commands' and then tell you to ask that 'INI MDI commands' are added to Gmoccapy if they were the same thing?HALUI MDI Commands = INI MDI Commands
Please Log in or Create an account to join the conversation.
02 Jan 2024 13:40 - 02 Jan 2024 13:42 #289558
by zz912
Replied by zz912 on topic Python Interface makes race conditions - mayby
To Aciera:
HALUI MDI commands
We think this:
github.com/LinuxCNC/linuxcnc/blob/2.9/src/emc/usr_intf/halui.cc
"Ask the Gmoccapy devs to add INI MDI command ability to Gmoccapy."
I don't like duplicating the same function.:
github.com/LinuxCNC/linuxcnc/blob/2.9/sr...gmoccapy/gmoccapy.py
I don't like duplicating the same function, but it would be easier for a programmer.
HALUI MDI commands
We think this:
github.com/LinuxCNC/linuxcnc/blob/2.9/src/emc/usr_intf/halui.cc
"Ask the Gmoccapy devs to add INI MDI command ability to Gmoccapy."
I don't like duplicating the same function.:
github.com/LinuxCNC/linuxcnc/blob/2.9/sr...gmoccapy/gmoccapy.py
I don't like duplicating the same function, but it would be easier for a programmer.
Last edit: 02 Jan 2024 13:42 by zz912.
Please Log in or Create an account to join the conversation.
02 Jan 2024 13:50 #289560
by zz912
Replied by zz912 on topic Python Interface makes race conditions - mayby
Variable for start and finish HALUI MDI Commands is here:
github.com/LinuxCNC/linuxcnc/blob/f7187c...r_intf/halui.cc#L253
I dont know how it read in Gmoccapy.py file. Unfortunately, this is only one of many problems. On the other hand, I think there are programmers who can read variables from c++ in python.
That's why I'm writing here a concept of how everything should work overall. I don't think it would be that complicated.
github.com/LinuxCNC/linuxcnc/blob/f7187c...r_intf/halui.cc#L253
I dont know how it read in Gmoccapy.py file. Unfortunately, this is only one of many problems. On the other hand, I think there are programmers who can read variables from c++ in python.
That's why I'm writing here a concept of how everything should work overall. I don't think it would be that complicated.
Please Log in or Create an account to join the conversation.
02 Jan 2024 13:55 #289562
by cmorley
Replied by cmorley on topic Python Interface makes race conditions - mayby
The difference is which program runs the MDI command. The GUI can't 'wait_complete' HALUI commands. You should ne able to avoid the race if the same program runs all MDI type commands.Cmorley:
I would also suggest not using HALUI MDI commands.
Ask the Gmoccapy devs to add INI MDI command ability to Gmoccapy.
zz912:HALUI MDI Commands = INI MDI Commands
Now I'm confused, why would Chris suggest NOT to use 'HALUI MDI commands' and then tell you to ask that 'INI MDI commands' are added to Gmoccapy if they were the same thing?
The following user(s) said Thank You: Aciera
Please Log in or Create an account to join the conversation.
02 Jan 2024 14:09 #289564
by Aciera
Replied by Aciera on topic Python Interface makes race conditions - mayby
Where does the python interface stand in all this?
If I have a remap with
' yield INTERP_EXECUTE_FINISH'
does that only wait for c.mdi() or 'self.execute()' commands of that particular remap to finish?
If I have a remap with
' yield INTERP_EXECUTE_FINISH'
does that only wait for c.mdi() or 'self.execute()' commands of that particular remap to finish?
Please Log in or Create an account to join the conversation.
02 Jan 2024 14:58 #289569
by cmorley
Replied by cmorley on topic Python Interface makes race conditions - mayby
Remap python is a different context..it's embedded not the same as the gui python.
I am talking of [Halui] MDI_COMMANDS=
They mostly work.
I am talking of [Halui] MDI_COMMANDS=
They mostly work.
The following user(s) said Thank You: Aciera
Please Log in or Create an account to join the conversation.
09 Jan 2024 13:27 #290270
by zz912
Replied by zz912 on topic Python Interface makes race conditions - mayby
Hello,
i was thinking about the problem with HALUI MDI_commands.
How it works:
1) Here it finds out what mode the LCNC was in before the commands start being executed:github.com/LinuxCNC/linuxcnc/blob/f7187c...halui.cc#L1027-L1031
2) HALUI MDI_commands are executed.
3) old mode is returninggithub.com/LinuxCNC/linuxcnc/blob/f7187c...halui.cc#L2131-L2141
There is usually a problem with restore mode. Python interface does not know that halui will change mode and start working.
I've been wondering all this time how to make the Python Interface know that halui is going to change modes.
But today a different philosophy occurred to me. If halui knows something is going on, it won't execute restore mode.
What do you think about it?
i was thinking about the problem with HALUI MDI_commands.
How it works:
1) Here it finds out what mode the LCNC was in before the commands start being executed:
if (!halui_sent_mdi) {
// There is currently no MDI command from halui executing, we're
// currently starting the first one. Record what the Task mode is,
// so we can restore it when all the MDI commands finish.
halui_old_mode = emcStatus->task.mode;2) HALUI MDI_commands are executed.
3) old mode is returning
if (halui_sent_mdi) { // we have an ongoing MDI command
if (emcStatus->status == 1) { //which seems to have finished
halui_sent_mdi = 0;
switch (halui_old_mode) {
case EMC_TASK_MODE_MANUAL: sendManual();break;
case EMC_TASK_MODE_MDI: break;
case EMC_TASK_MODE_AUTO: sendAuto();break;
default: sendManual();break;
}
}
}There is usually a problem with restore mode. Python interface does not know that halui will change mode and start working.
I've been wondering all this time how to make the Python Interface know that halui is going to change modes.
But today a different philosophy occurred to me. If halui knows something is going on, it won't execute restore mode.
if (halui_sent_mdi) { // we have an ongoing MDI command
if (emcStatus->status == 1) { //which seems to have finished
halui_sent_mdi = 0;
if (emcStatus->task.interpState == EMC_TASK_INTERP_IDLE) {
switch (halui_old_mode) {
case EMC_TASK_MODE_MANUAL: sendManual();break;
case EMC_TASK_MODE_MDI: break;
case EMC_TASK_MODE_AUTO: sendAuto();break;
default: sendManual();break;
}
}
}
}What do you think about it?
Please Log in or Create an account to join the conversation.
10 Jan 2024 06:14 #290329
by Aciera
Replied by Aciera on topic Python Interface makes race conditions - mayby
if it solves the issues you are having and does not seem to break anything then maybe file this as a pull request on github. Otherwise it is likely to just get buried here.
The following user(s) said Thank You: zz912
Please Log in or Create an account to join the conversation.
10 Jan 2024 07:26 #290331
by zz912
For example, there are many of my theories in which I contradict myself.
github.com/LinuxCNC/linuxcnc/issues/2453
That's why I like to put only what I'm at least somewhat sure about on Github.
I also contradict myself in this thread. First I started with theories where I would like to add signals to check if a process is already running. After that I thought about it and now I think it could be solved via IDLE mode. IDLE mode is stuff that is and has always been part of LCNC, so I wouldn't complicate anything.
For example, Issue 2453 is very insidious in that some developers are unable to simulate it at all or only to a very limited extent to be able to fix it.
Therefore, it is necessary to go through the code manually and remove situations where race conditions could arise.
The problem is that a problem can be caused by several situations, so if we remove only one, nothing may appear, but it is necessary to proceed in this way if we want LCNC to be a stable system.
Replied by zz912 on topic Python Interface makes race conditions - mayby
You don't have to worry about that. I will definitely do a PR. Unfortunately, I've already made a lot of mess on Github with my theories.Otherwise it is likely to just get buried here.
For example, there are many of my theories in which I contradict myself.
github.com/LinuxCNC/linuxcnc/issues/2453
That's why I like to put only what I'm at least somewhat sure about on Github.
I also contradict myself in this thread. First I started with theories where I would like to add signals to check if a process is already running. After that I thought about it and now I think it could be solved via IDLE mode. IDLE mode is stuff that is and has always been part of LCNC, so I wouldn't complicate anything.
For example, Issue 2453 is very insidious in that some developers are unable to simulate it at all or only to a very limited extent to be able to fix it.
Therefore, it is necessary to go through the code manually and remove situations where race conditions could arise.
The problem is that a problem can be caused by several situations, so if we remove only one, nothing may appear, but it is necessary to proceed in this way if we want LCNC to be a stable system.
Please Log in or Create an account to join the conversation.
10 Jan 2024 08:52 #290341
by Aciera
Replied by Aciera on topic Python Interface makes race conditions - mayby
No problem.
If your unhappy with an issue or pull-request you made on github then just close it to tidy things up.
I usually need several attempts with pull requests and issues myself.
If your unhappy with an issue or pull-request you made on github then just close it to tidy things up.
I usually need several attempts with pull requests and issues myself.
Please Log in or Create an account to join the conversation.
Time to create page: 0.141 seconds