Multi input AND/OR; by dirkx · Pull Request #4720 · LinuxCNC/linuxcnc

dirkx · 2026-10-24T21:43:17Z

Multi input AND/OR; much along the lines of logic(9) but with a controlled unconnected pin value that helps during machine conversation (as you can simply comment out some feed) and an easier to read name.

Should implement all @BsAtHome and others their suggestions in issue #4711 and #4721. Manual pages converted to adoc.

…olled unconnected pin value that helps during machine conversation and an easier to read name.

BsAtHome · 2026-10-24T22:06:23Z

+option default_personality 8;
+option default_count 1;
+
+license "ASLv2";


What type of license is that?

It would probably fail for RTAI because anything not GPL will taint the kernel.

Besides, if you need this specific license, then it must also be added to debian/copyright, which needs consensus to ensure compatibility.

changed to GPL; Was the Apache Software License v2

BsAtHome · 2026-10-24T22:08:10Z

@@ -0,0 +1,36 @@
+component and "Multi input AND gate";
+pin in bool in##[128:personality] = 1 "input pins, up to personality-1";


Is 128 not excessive?

Can you give an example where you'd need more than 40..50 inputs AND'ed together?

So the case I have here/that started here was steps control in a 2D conveyer belt/palletrizer is 4x5 position = 20 pallets x (8 mount sensors) + 5 normal interlocks. So I rounded up to the next multiple of 2. Happy to make that 64.

The other case is a pneumatic sorter where I have two sets(in/out) of blockage/low-pressure/high pressure detects in each of the 41 (don't ask -- I suspect the designer mixed inches v.s metric) linear stations. Though in this case you could roll up each of the 2x3 sets into a 41 OR and then combine each pair with a normal OR into the 3 outputs needed for the pump/tank.

Convinced. I guess it would probably be better to use 255 instead of 128. That would give pins in-00 through in-99 and then three digits up til in-254.

The reason for not having this at 256 is that logic uses the last byte (does personality & 0xff). That effectively creates pins 0..254 on the outset in logic.

BsAtHome · 2026-10-24T22:09:37Z

@@ -0,0 +1,33 @@
+component or "Arbitrary input OR gate";
+pin in bool in##[128:personality] = 0 "in";


The doc string "in" is a bit thin, isn't it?

Grin :) took it from one of the other modules - elaborated :)

Ok - was not able to remove it - as a "" or none will give corrupted output / seems to concatenate the in and out. So have instead made it a more elaborate string.

BsAtHome · 2026-10-24T22:11:37Z

@@ -0,0 +1,33 @@
+component or "Arbitrary input OR gate";


The AND gate you use the term "Multi". Why is this "Arbitrary"? What is arbitrary about these inputs?

Why not retain the comp doc string I wrote?

BsAtHome · 2026-10-24T22:14:38Z

@@ -0,0 +1,36 @@
+component and "Multi input AND gate";


The component naming and and or may prove confusing with and2 and or2. That is why I called them and_n and or_n. The underscore is converted into a dash for the hal internals, but if you don't like that, then you can use names=.

So, any particular reason to the naming?

None particularly; but _n seeded a bit extra. Changed to _n.

BsAtHome · 2026-10-24T22:15:38Z

+function _ "Logical AND";
+
+see_also """
+*or*(9),


Please don't use emphasis. The post-processor cannot handle it and links will no longer be provided in the html version.

Grin :) took it from one of the other modules; now removed.

BsAtHome · 2026-10-24T22:16:25Z

+
+option period no;
+option default_personality 8;
+option default_count 1;


The default count is implicitly one. Why set it?

Grin :) took it from one of the other modules - removed :)

BsAtHome · 2026-10-24T22:19:00Z

+is TRUE.
+""";
+
+function _ "Logical AND";


Here you add a doc string. But in the or you don't. Not consistent. You may also want to elaborate... I think it is rather obvious that it calculates the logical AND. But what makes this so special, that is why you document it.

ok - will fix by removing this for the AND

andypugh · 2026-10-25T20:36:21Z

If the main improvement over "logic" is that this one allows for a default value for unconnected inputs, would it make more sense to modify "logic" to support this?

dirkx · 2026-10-25T20:57:11Z

That was exaxtly my first approach - the problem is that this gets quite complex/weird for things like an xor/xand and so on; you almost have to introduce third HiZ style state (which makes things slow as you need to do it every loop - as halcmd can chance something).

And given how trivial and AND and OR are; and how common these are (lots of stations on a line all needing to request hydraulics or air and this OR-ed together for the main pump; or lots of interlocks getting combined) -- this was easy/natural and also makes the hal files easy to read/understand.

Obviously the other approach would be to generate more of the HAL files with some m4 or similar macro language; and make the numbering/creation much more automatic/generated. But that makes the CE/machine directive and other paperwork a lot more complex. So hence this strawman.

rene-dev · 2026-10-25T23:00:28Z

would these replace and2 and or2?

grandixximo · 2026-10-26T04:24:00Z

would these replace and2 and or2?

I don't think so, these seems to be useful for when you have to group together alarm signals or similar into one single output, but and2/or2 are much more versatile than that.

BsAtHome · 2026-10-29T09:03:29Z

@andypugh Why was this merged? I had a lot of comments that were replied upon but never changed in the code.

andypugh · 2026-10-29T13:31:20Z

@andypugh Why was this merged? I had a lot of comments that were replied upon but never changed in the code.

I mis-read and thought that the issues had been addressed.

andypugh · 2026-10-29T13:34:20Z

It does not appear to be possible to re-open the PR, which is going to complicate matters.

grandixximo · 2026-10-29T13:36:48Z

@dirkx are you able to work on the review pointers, and send in a new PR?

BsAtHome · 2026-10-29T13:37:04Z

It does not appear to be possible to re-open the PR, which is going to complicate matters.

Yeah, that is a bummer...

@dirkx you should re-submit the PR with the suggested changes, which you also acknowledged. Then it will run through review once again.

dirkx · 2026-10-29T19:22:02Z

Ack. Done as #4746.

Multi input AND/OR; much along the lines of logic(9) but with a contr…

d019900

…olled unconnected pin value that helps during machine conversation and an easier to read name.

BsAtHome reviewed Oct 24, 2026

View reviewed changes

andypugh merged commit 536d1ba into LinuxCNC:master Oct 28, 2026
17 checks passed

dirkx mentioned this pull request Oct 29, 2026

Add generic and/or with flexibel inputs & allow unconnected pins #4746

Open

		@@ -0,0 +1,36 @@
		component and "Multi input AND gate";
		pin in bool in##[128:personality] = 1 "input pins, up to personality-1";

		@@ -0,0 +1,33 @@
		component or "Arbitrary input OR gate";
		pin in bool in##[128:personality] = 0 "in";

Conversation

dirkx commented Oct 24, 2026

Uh oh!

Choose a reason for hiding this comment

Uh oh!

dirkx Oct 25, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

dirkx Oct 25, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

andypugh commented Oct 25, 2026

Uh oh!

dirkx commented Oct 25, 2026

Uh oh!

rene-dev commented Oct 25, 2026

Uh oh!

grandixximo commented Oct 26, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

BsAtHome commented Oct 29, 2026

Uh oh!

andypugh commented Oct 29, 2026

Uh oh!

andypugh commented Oct 29, 2026

Uh oh!

grandixximo commented Oct 29, 2026

Uh oh!

BsAtHome commented Oct 29, 2026

Uh oh!

dirkx commented Oct 29, 2026

Uh oh!

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

dirkx Oct 25, 2026 •

edited

Loading

dirkx Oct 25, 2026 •

edited

Loading

grandixximo commented Oct 26, 2026 •

edited

Loading