Add method to take with message_info #542
Conversation
Signed-off-by: Luetkebohle Ingo (CR/AEX3) <ingo.luetkebohle@de.bosch.com>
Signed-off-by: Luetkebohle Ingo (CR/AEX3) <ingo.luetkebohle@de.bosch.com>
Signed-off-by: Luetkebohle Ingo (CR/AEX3) <ingo.luetkebohle@de.bosch.com>
|
I've reworked this to take comments into acount, but we still have several test failures that I can't really figure out right now. I'll look into it tomorrow. |
Signed-off-by: Luetkebohle Ingo (CR/AEX3) <ingo.luetkebohle@de.bosch.com>
|
OK, I fixed the tests. It was a really simple issue because of a bad merge. btw, since I now integrated taking the message-info with the "normal" rcl_take, we can be sure it won't segfault at least. the correctness of the timestamps is not yet checked, because none of the existing tests can be easily modified, I will have to write a new one. |
|
I really think we need a basic test to show how it would be used in python and to ensure that the timestamps are getting converted correctly (whether they be 0 or some value). Do you have time to work on that still? I can start testing in the meantime. |
When is the deadline? If it has to be today, that'd be tough. It's already almost 9pm here and I'm also looking at the service timestamps. |
|
Today is the deadline. |
|
Alright. I'll get it done. |
|
There are also some compiler warnings on Windows: https://ci.ros2.org/job/ci_windows/10284/msbuild/new/ If you need to get off, that's fine, just tell me what's not done and I'll do my best to finish things. |
Thanks for the offer. So far, I'm planning to continue for a while, probably at least one more hour or maybe two. Those warnings look easy, I'll get to them. |
Signed-off-by: Luetkebohle Ingo (CR/AEX3) <ingo.luetkebohle@de.bosch.com>
|
I have now added a test. It is really very basic, but should do the trick. |
Signed-off-by: Luetkebohle Ingo (CR/AEX3) <ingo.luetkebohle@de.bosch.com>
|
@wjwwood okay, with the last commit this should now also solve the warning problem on Windows. |
|
Yup I’m going to rerun ci and stuff ASAP. |
Signed-off-by: Luetkebohle Ingo (CR/AEX3) <ingo.luetkebohle@de.bosch.com>
|
🥳 |
| result = _rclpy.rclpy_take(capsule, sub.msg_type, False) | ||
| if result is not None: | ||
| msg, info = result | ||
| self.assertNotEqual(0, info['source_timestamp']) |
There was a problem hiding this comment.
This test always fails with Connext: see #615.
This implements the Python side of ros2/design#259 and needs ros2/rcl#619
The message info is simply returned as a dictionary. I figured if we want to encapsulate it into something else, we can always do that in Python. That shouldn't break API.
So far, only subscriptions have a method to take with info.