Support ZeroMQ PUB/SUB endpoints

[m3d]

PUB/SUB modes are useful for remote control operations. Moreover based on the team experience actually Pull/Push is not recommended in this scenario as the receiver is not able to easily control the throughput.

I was not really sure if to create 2 new OSGAR drivers with one line difference of type selection. Would you go for this self.use_pubsub_mode = config.get("pubsub", False) "hack" or what would be the recommendation?

I will also start PR in osgar-apps as user of this functionality for remote control with streaming video.

[zbynekwinkler]

Don't you need some changes also at other places? Sending, receiving...

[m3d]

so far it looked like that all is needed is this extra: socket.setsockopt_string(zmq.SUBSCRIBE, "") ... but to be honest I did not test this change on the robot

…

On Fri, Jan 30, 2026 at 4:18 PM Zbyněk Winkler ***@***.***> wrote: *zbynekwinkler* left a comment (robotika/osgar#1037) <#1037 (comment)> Don't you need some changes also at other places? Sending, receiving... — Reply to this email directly, view it on GitHub <#1037 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABLG4YPYDUR4NRDDPDEFZST4JNY5RAVCNFSM6AAAAACTJ4ESB6VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTQMRUGI3TKOBQG4> . You are receiving this because you authored the thread.Message ID: ***@***.***>

[zbynekwinkler]

Well, I am sure the send a receive calls need/return multipart messages that deal with subscriptions.

[tajgr]

It looks save regarding to old features. I am not sure if I can test the new mode now. Hopefully, I will try something during weekend.
Otherwise, I agree with the expansion of pull and push.
Are you in a hurry for this PR?

[m3d]

OK, thanks - no hurry. I plan to do the 2nd round of tests next week.

…

On Fri, Jan 30, 2026 at 10:23 PM Jakub Lev ***@***.***> wrote: *tajgr* left a comment (robotika/osgar#1037) <#1037 (comment)> It looks save regarding to old features. I am not sure if I can test the new mode now. Hopefully, I will try something during weekend. Otherwise, I agree with the expansion of pull and push. Are you in a hurry for this PR? — Reply to this email directly, view it on GitHub <#1037 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABLG4YP2YOWNEWWHV5IOFQ34JPDVNAVCNFSM6AAAAACTJ4ESB6VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTQMRVHAYTMOJWGM> . You are receiving this because you authored the thread.Message ID: ***@***.***>

[tajgr]

I used this older sample for test and it worked fine.

[m3d]

I used this older sample for test and it worked fine.

OK, but be aware that if you do zmq.SUB you also need to add one more call with zmq.SUBSCRIBE similarly to what is in PR. Thanks

[m3d]

p.s. the test with remote control on robot works fine ...

[m3d]

note, that the FAIL: test_environ (osgar.test_logger.LoggerStreamingTest) is also on master. I actually see the warning, but it is in RED (?!) and I guess this was not before?? Can it have an influence? On the other hand if I run just the test itself it does not print the message and passes

...............Environment variable OSGAR_LOGS is not set - using working directory
F...

(python) md@md-ThinkPad-P50:~/git/osgar$ python test.py osgar/test_logger.py
...........................
----------------------------------------------------------------------
Ran 27 tests in 0.332s

OK

[m3d]

the problem with logging is caused by np.testing.assert_almost_equal(mat[:2, :2], np.identity(2)) ... i.e. numpy.testing

[m3d]

OK, I do not see any complains, so merging ...