[REP-2011] Evolving Message Types #358

wjwwood · 2022-06-10T01:12:50Z

This REP (tentatively REP-2011) proposes patterns and approaches for evolving message, service, and action ROS interface types over time.

See the draft in the changes of this pr for information.

Thanks to @methylDragon, @youliangtan, @kylemarcey, and others for their help on this REP.

Status

Generally the REP is complete enough to invite technical feedback and discussions. Where things are poorly described, rationalized, or a particular alternative isn't listed, please point it out, but keep in mind that we may just haven't gotten to it yet.

REP writing by section:
- Motivation: done
- Specification: done, with TODOs
- Rationale: partially done, with TODOs and a few places that need to be expanded
- Backwards Compatibility: not started
- Feature Progress: not started
Specification Reference Implementation:
- Type Version Enforcement: not started
- Type Description Distribution: not started
- Run-Time Interface Reflection: not started
- Tools for Evolving Types: not started

Known Issues

The REP needs to be updated to consider Services and Actions in several places
The parts about TypeDescription should be factored out into a separate section and referenced from the Type Version Enforcement, Type Description Distribution, and Run-Time Interface Reflection sections
The type version hash should probably be based on the TypeDescription, but the question about type name being part of the hash needs to be resolved first
The Run-Time Interface Reflection section should be expanded or split into the part about reflecting on serialized buffers using a TypeDescription and the part about creating "type support" structs for creating publishers and subscriptions from a TypeDescription
It is not yet clear how we will use the transfer functions for Services. It could be as a relay node like topics, or as plugins that are colocated with the Service client or server
There are some possibly upcoming changes to what is supported from .idl files, e.g. like enums, which may impact the TypeDescription type, and more generally we might want to consider how future changes to this type might be mitigated, e.g. by coming up with a way to handle .idl annotations in the type
We need a way to represent the "user-defined interface version" in either .idl or .msg files or both, or otherwise remove the idea from the REP, though it would be nice to preserve it
And many other smaller todo items left in the REP

Signed-off-by: William Woodall <william@osrfoundation.org>

Signed-off-by: methylDragon <methylDragon@gmail.com>

Signed-off-by: methylDragon <methylDragon@gmail.com> Co-authored-by: William Woodall <william+github@osrfoundation.org>

* Increase readability, fix typos, and reduce ambiguities Signed-off-by: methylDragon <methylDragon@gmail.com> * Add clarification for field semantics Signed-off-by: methylDragon <methylDragon@gmail.com> * Add first draft of inter-process type description distribution Signed-off-by: methylDragon <methylDragon@gmail.com> * Add additional notes for tick-tocking the metatype Signed-off-by: methylDragon <methylDragon@gmail.com> * Add rationale section for inter-process type description distribution Signed-off-by: methylDragon <methylDragon@gmail.com> * Add additional note about sending type description metadata Signed-off-by: methylDragon <methylDragon@gmail.com> * Add alternatives section for type description distribution Signed-off-by: methylDragon <methylDragon@gmail.com> * Fix broken TODO:: directives Signed-off-by: methylDragon <methylDragon@gmail.com> * Apply rewording suggestions for type description distribution Signed-off-by: methylDragon <methylDragon@gmail.com> Co-authored-by: William Woodall <william+github@osrfoundation.org> * Apply rewording suggestions from code review Signed-off-by: methylDragon <methylDragon@gmail.com> Co-authored-by: William Woodall <william+github@osrfoundation.org> Co-authored-by: William Woodall <william+github@osrfoundation.org> * remove trailing whitespace Signed-off-by: William Woodall <william@osrfoundation.org> * fixups while walking through it with Brandon Signed-off-by: William Woodall <william@osrfoundation.org> Co-authored-by: methylDragon <methylDragon@gmail.com>

Signed-off-by: William Woodall <william@osrfoundation.org>

* Add section on runtime type introspection Signed-off-by: methylDragon <methylDragon@gmail.com> * Apply REP rephrases Signed-off-by: methylDragon <methylDragon@gmail.com> Co-authored-by: William Woodall <william+github@osrfoundation.org> * Apply more REP rephrases Signed-off-by: methylDragon <methylDragon@gmail.com> Co-authored-by: William Woodall <william+github@osrfoundation.org>

* Add nested type description section Signed-off-by: methylDragon <methylDragon@gmail.com> * Use Field type in message descriptions Signed-off-by: methylDragon <methylDragon@gmail.com> * Fix unmatched parenthesis Signed-off-by: methylDragon <methylDragon@gmail.com> * Implement nested introspection API Signed-off-by: methylDragon <methylDragon@gmail.com> * Rework nested introspection API Signed-off-by: methylDragon <methylDragon@gmail.com>

Signed-off-by: youliang <tan_you_liang@hotmail.com>

Signed-off-by: William Woodall <william@osrfoundation.org>

ros-discourse · 2022-08-22T06:13:48Z

This pull request has been mentioned on ROS Discourse. There might be relevant details there:

https://discourse.ros.org/t/rfc-rep-2011-evolving-message-types/27006/1

james-rms · 2022-08-23T22:43:48Z

Hello! Glad this is moving along.

Do you have any ideas or vision around how the Type Description would be serialized?
Do you have any ideas about how ros2 bag play would provide this type description for topics recorded using older types?

wjwwood · 2022-08-24T03:33:10Z

Do you have any ideas or vision around how the Type Description would be serialized?

Sure, as described in the "Type Description Distribution" section, it will be a ROS message, called TypeDescription.msg (probably). So it will be serialized using the chosen rmw implementation. Inside the bag file it may be serialized as the rmw implementation would (e.g. CDR) or we could store it in a different format (e.g. json or yaml) if we wanted. We can also store the original description text and use that to recreate the TypeDescription instance.

Do you have any ideas about how ros2 bag play would provide this type description for topics recorded using older types?

Yeah, the idea of how it would work goes like this:

when doing ros2 bag record rosbag will save the TypeDescription instance and the original description text from the .msg/.idl file into the bag file by calling the ~/_get_type_description service on the node that is publishing the data
when doing ros2 bag play, the TypeDescription instance will be retrieved from the bag file and used to create a publisher
creating the publisher will give the node the type description at the version that was recorded into the bag originally

Anyone asking the rosbag node about that type through the ~/_get_type_description service will get the original description from the bag file as that's what was used to create the publisher.

For bag files that were created before this feature is implemented, we may be able to have a way to "augment" the bag file during playback, specifying a file which contains a copy of the type description text for the types at those versions in the bag file.
You would have to go back and get that information some how, but you could do that from the git history in most cases.

vmayoral

As a general comment, this is a great addition to ROS 2 and a fantastic REP proposal. Very well explained. I wish we could've had this long ago, as it would have saved much trouble in past projects.

I dropped some comments and questions below.

vmayoral · 2022-08-24T07:13:25Z

rep-2011.rst

+Abstract
+========
+
+This REP proposes patterns and approaches for evolving message, service, and action ROS interface types over time.


Take this as the input of a non-native speaker but I miss in here a bit of a short-ish "what for" in the abstract. Evolving message, service, and action ROS interface types over time is not self-explanatory. I know that the motivation section comes right afterwards but still, an abstract should land this IMHO. I read it a few times and still missed it. Maybe something along these lines:

Suggested change

This REP proposes patterns and approaches for evolving message, service, and action ROS interface types over time.

This REP proposes patterns and approaches for evolving message, service, and action ROS interface types over time. Such capability can be used for evolving ROS 2 itself over time while addressing interoperability issues due to mistmaching type definitions.

As a side observation, AFAIK, when compared to other REPs, the abstract in here is significantly larger.

The Motivation section is well reasoned.

rep-2011.rst

vmayoral · 2022-08-24T08:17:46Z

rep-2011.rst

+
+    sensor_msgs/msg/Image__RIHS1_XXXXXXXXXXXXXXXXXXXX
+
+This has the benefit of "just working" for most middlewares which at least match based on the name of the type, and it is simple, requiring no further custom hooks into the middleware's discovery or matchmaking process.


Any consideration to be made on the max. lengths of the type names? Any restrictions to consider? Given the concatenated type version hash there might be little margin for using namespaces (which larger graphs tend to use).

That's a good point, I suppose we'd have to consider the DDSI-RTPS spec that limits topic lengths to 256 characters...

Though, looking at what such a character length even looks like:

sensor_msgs/msg/Image__RIHS1_XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX

I think there's plenty of space for multiple nested namespaces? I'd imagine the type version hash wouldn't need more than 32 characters to have a relatively low chance of collision, but that's just a guess off the top of my head.

In which case a topic name would have 224 characters, which is still a lot.

sensor_msgs/msg/Image__RIHS1_XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX

Also, since we're speaking about type names here (not topic names), the convention is to use <package name>/<type, msg/srv/action>/<name>, and so deeply nested namespaces isn't really as big of an issue as it is with things like topic names and node names.

In table 9.12 of the DDSI-RTPS 2.2 spec (https://www.omg.org/spec/DDSI-RTPS/2.2/PDF) it says the type name is limited to 256 characters as well. So the limit is the same, but again, I don't think we're in danger here of running into this limitation in practical situations. At least not very often.

One exception to this is the feature that allows you to configure topic endpoint's QoS using parameters: http://design.ros2.org/articles/qos_configurability.html

In this case the topic name is part of the parameter name, but unlike topic/service/node names, we don't have a limit for the length of parameter names at the moment.

vmayoral · 2022-08-24T08:33:17Z

rep-2011.rst

+It should be along side the ``topic_type`` string in that struct, but the ``topic_type`` field should not include the concatenated type version hash, even if the recommended approach is used.
+Instead the type name and version hash should be separated and placed in the fields separately.
+
+This information should be transmitted as part of the discovery process.


I'm not sure if I understand properly the meaning of the "discovery process" in here.

To my understanding, generally, for most DDS implementations, discovery traffic is sent prior/outside of the security plugins domain (i.e. this information can be used for reconnaissance). If this information is sent also as part of the DDS discovery process, won't it mean that we'd be conveying it in an uncrypted manner (regardless of whether security is enabled or not in the graph)?

Also, (a final answer would of course depend on the implementation but) can you briefly comment on what's the expectation regarding additional traffic generated due to sending this information as part of discovery? Will it be just the type version hash for each ROS interface?

Well, topic names are already sent as part of the rmw_topic_endpoint_info_t struct. Adding a hash in my opinion is not much of a security risk, since the hash would be degenerate, and by design, meant to not be reversible.

Furthermore, depending on how the hash is generated (e.g. factoring in the field names in the type description), it can be made more robust to rainbow table attacks (definitely more than something like, say, brute forcing of MD5 hashes for passwords.)

Regarding additional traffic, I'd think it'd just be the type version hash. The rest of the stuff (e.g. type description distribution) is planned to be sent after discovery, after the participants have determined that they're mismatched via the hash on discovery.

@methylDragon is right, it should just be the version hash, which would be in addition to the type name which already being sent via discovery. So the impact to information leak and extra bandwidth used should be minimal. The hash is going to be a sha-1 of the type description struct (most likely) so that is reversible, but not easily and not for much gain I think.

The actual full type descriptions (and other meta data) would be sent via the ~/_get_type_description service which will have to be considered in security set ups just all the other built-in node topics/services must.

james-rms · 2022-08-29T22:23:16Z

rep-2011.rst

+Type Description Contents and Format
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+The response sent by the ROS Service server will contain a combination of the original ``idl`` or ``msg`` file's content, as well as any necessary information to serialize and deserialize the raw message buffers sent on the topic.


It's probably an important detail that the full idl or msg definition of all field types needs to be provided along with the toplevel definition, otherwise there may be an unresolved incompatibility between types at a field level. It's worth noting that ROS1 tries to solve this by concatenating all of its msg definitions using gendeps --cat and sending this in the connection header for new connections.

Yup! You're right! They're meant to be sent along the main type description under the referenced_type_descriptions field.

At least, for any non-primitive types.

Do you see value in concatenating the referenced file contents as well (including the comments)? Or would a parsed definition suffice?

Do you see value in concatenating the referenced file contents as well (including the comments)? Or would a parsed definition suffice?

It depends on how frozen-in-time the parsing procedure is. If the parsed definition evolves over time, I'd keep the original file content around, because then if there's a Type Description schema mismatch, a node can fall back to rebuilding the type description from the original file content. Stripping comments seems safe enough, since at least then the content is still valid IDL or MSG.

@wjwwood regarding storing a type definition source (msg or idl) as a string, I wanted to call your attention to work that has been happening over in mcap.

The specific problem I am referencing is how to store both the message definition and all direct and transitive dependencies in the proposed type_description_raw field. ROS 1 solves this using the gendeps --cat format (mentioned by @james-rms above), which consists of msg files delimited by 80 = characters plus a message name line.

For storing concatenated source message definitions in MCAP, we opted to follow the ROS 1 convention for msg files, which we use for both ROS 1 and ROS 2 data. For IDL source message definitions, we modified the concatenation slightly so that all types including the top level type are explicitly named.

You can see our specification for this here, and discussion on the spec PR at foxglove/mcap#553. This spec was then implemented in ros-tooling/rosbag2_storage_mcap#53.

I think it would be best if we can use this same concatenation standard for REP 2011 types.

It was pointed out to me that you could avoid concatenating message definition strings entirely by providing an array of types (the top level type and any dependent types, each with a name, format, and string body).

Signed-off-by: William Woodall <william@osrfoundation.org>

ros-discourse · 2022-09-20T00:46:27Z

This pull request has been mentioned on ROS Discourse. There might be relevant details there:

https://discourse.ros.org/t/ros-2-tsc-meeting-minutes-2022-9-15/27394/1

ros-discourse · 2022-11-09T23:29:24Z

This pull request has been mentioned on ROS Discourse. There might be relevant details there:

https://discourse.ros.org/t/ros-2-tsc-meeting-minutes-2022-10-13/28213/1

fujitatomoya · 2023-01-06T23:57:44Z

rep-2011.rst

+
+.. TODO::
+
+    discuss the implications for large messages and the possibility of having the transfer functions be colocated with either the publisher or subscription more directly than with component nodes and remapping.


probably this depends on version mismatch situation. if each subscription does have different message version, that would be efficient that transfer functions should be processed on subscription, since it will delay the publisher to send the message out to the network.
on the other hand, and i think this is likely that maybe 1 or 2 version mismatch and multiple subscriptions in the topology. in that case, transfer functions should be processed in publisher side to avoid to call transfer function on each subscription.

besides, this would be nice to configured? because it depends on the platform resource, no matter what mismatch situation it is, platform resource could be the problem for this extra transfer function, thinking about far edge devices.

fujitatomoya · 2023-01-07T00:03:28Z

rep-2011.rst

+
+These features are described in the following sections.
+
+Type Version Enforcement


It is more like detection as feature? user must be able to know if the type version mismatches or not, then it is up to user application to enforce and transfer the version?

emersonknapp

Some "my 2 cents" mostly on TODO items, couple questions, clarifications

emersonknapp · 2023-01-10T07:06:51Z

rep-2011.rst

+
+.. TODO::
+
+    Related TODO, should we just use the TypeDescription described in a later section?


I think yes. Has there been further discussion on this point, as to whether the has should be based on anything other than the TypeDescription? It seems like TypeDescription peels away the unrelated info (comments, default values, etc.) leaving only the relevant information to hash.

emersonknapp · 2023-01-10T07:09:13Z

rep-2011.rst

+
+.. TODO::
+
+    Remove the idea of the user-defined interface version or define how it can be supplied by the user in one or more of the idl file kinds.


Perhaps just allow users to version via a recommended strategy of having a "versionN: true" field that is replaced for new versions? First thought was have it have a default value, or define a const, but I think based on current description of hashing logic, those two particular things should not change the hash. Inclined to agree that we don't need a special mechanism in this REP for "manual hash break".

emersonknapp · 2023-01-10T07:16:14Z

rep-2011.rst

+
+.. code::
+
+    sensor_msgs/msg/Image__RIHS1_XXXXXXXXXXXXXXXXXXXX


Small concern here, maybe just due to my bad memory - do we require all RMW impls to use the provided QoS profiles? If so, and we are planning to add USER_DATA to provide type hash, then do we need this backup strategy of concatenated type name?

If USER_DATA not universally viable - if one does use this strategy, is the plaintext typename totally redundant (because type name is in the TypeDescription object that produces the hash, and so prevents hash collision for types even of the same layout with different names)? If redundant, would it be better to just advertise type RIHS_1_xxx? Maybe not, since hash is not reversible, and the prefix allows for checking for transfer functions. I may have not been thinking of the whole scope of the REP.

Not sure about QoS profiles.

But for advertised type names.. Putting the type name as a prefix in the advertised type will allow us to bypass querying for the TypeDescription object in the case where we just want the name of the type the participant is trying to use (e.g. in a tool to see what participants are trying to communicate using the same type (the prefix), but can't because of the layout (different hashes))

emersonknapp · 2023-01-10T07:20:19Z

rep-2011.rst

+
+.. TODO::
+
+    It's not clear how we will do this just now, since existing "QoS events" lack a way to communicate this information, I (wjwwood) think.


QoS events lack a way to communicate which information? The specific endpoint? The current events, provided in callbacks, do carry event-specific struct payloads that are filled out by the RMW implementation. The current ones are all simple number types at the moment, but that's not a hard requirement. I don't see a reason why rmw_topic_endpoint_info_t couldn't be included in the callback payload even, or just the GID or something from it.

Noting that this callback type is now implemented, so this TODO (and my prior comment) are resolved

emersonknapp · 2023-01-10T07:22:58Z

rep-2011.rst

+
+.. TODO::
+
+    How can we detect when a remote node is not offering this service?


Just agreeing that yes this is challenging. rosbag2 is going to be a first adopter of these services, and have a strong preference for finding type descriptions. How will it know when to stop checking? Like you've pointed out, hard to tell whether not ready yet, or never will be.

emersonknapp · 2023-01-10T07:25:40Z

rep-2011.rst

+
+.. TODO::
+
+    Should each endpoint be asked about their type/version pair or should we assume that the type/version pair guarantees a unique type description and therefore reuse past queries?


It seems to me best to make a strong statement that type/version pair is a unique type description so a user can expect that anybody with that name and hash has the same type. From bag recording perspective, with potentially a lot of topics, it'd be nice to only call out once per unique type, and I'd probably just base it on hash lookup in internal definition cache, whether to call or not

emersonknapp · 2023-01-10T07:27:34Z

rep-2011.rst

+.. TODO::
+
+    What happens if the message consumer doesn't have access to the serialization library stated in the meta-type?
+    (wjwwood) It depends on what you're doing with the response. If you are going to subscribe to a remote endpoint, then I think we just refuse to create the subscription, as communication will not work, and there's a basic assumption that if you're going to communicate with an endpoint then you are using the same or compatible rmw implementations, which includes the serialization technology. The purpose of this section and ROS Service is to provide the information, not to ensure communication can happen.


I think it's fair to say that cross-protocol communications is out of scope for this REP, even if we are providing information that could unlock it down the line.

Agreed. That said, I think this part of the discussion should be preserved, possibly by moving this paragraph out of the TODO section and doing some editing.

emersonknapp · 2023-01-10T07:28:38Z

rep-2011.rst

+
+.. TODO::
+
+    (wjwwood) I propose we use key-value string pairs for the extra information, but I am hoping for discussion on this point.


Maybe I missed it, but what extra_information are we anticipating here? Not sure what the value of debating representation is, without a sense of what it's for.

I think this is more used as a final catch-all to mitigate the case where a future feature/bug getting uncovered forces us to release a new version of the type description message

emersonknapp · 2023-01-10T07:31:10Z

rep-2011.rst

+
+.. TODO::
+
+    (wjwwood) A thought just occurred to me, which is that we could maybe have "lazy" Service Servers, which watch for any Service Clients to come up on their Service name and when they see it, they could start a Service Server.


What overhead is presented by a ServiceServer? Unlike a Publisher, it doesn't do work unless called, so I'm not seeing a comparison to the lazy publisher. Is it discovery overhead that's the main concern, or is there also a non-negligible memory/cpu contribution to creating the server? How does that overhead compare to the monitoring required to see whether there are any prospective Clients?

emersonknapp · 2023-01-10T07:31:47Z

rep-2011.rst

+
+    This section needs to be updated to be inclusive to at least Services, and then we can also mention Actions, though they are a combination of Topics and Services and so don't need special support probably.
+
+Plugin Interface


Nit: should this be titled "Serialization Library Plugin Interface"?

clalancette

I realize that I never left my actual review. I have two comments inline so far, one of which is already addresses in a PR to this PR.

clalancette · 2023-01-10T14:47:24Z

rep-2011.rst

+    It's essentially the same thing, but it does include the type name.
+    I (wjwwood) am leaning in this direction.
+
+The resulting data structure is hashed using a standard SHA-1 method, resulting in a standard 160-bit (20-byte) hash value which is also generally known as a "message digest".


At the risk of opening a can of worms, I think we should consider whether we really want to use SHA-1 here. While I understand this isn't meant to be used cryptographically, all of the signs seem to say that we should be migrating away from SHA-1 now. So unless we absolutely have to use SHA-1 for some kind of compatibility, I feel like we should choose a different hashing algorithm (maybe SHA-256).

clalancette · 2023-01-10T15:18:10Z

rep-2011.rst

+.. TODO::
+
+    What happens if the message consumer doesn't have access to the serialization library stated in the meta-type?
+    (wjwwood) It depends on what you're doing with the response. If you are going to subscribe to a remote endpoint, then I think we just refuse to create the subscription, as communication will not work, and there's a basic assumption that if you're going to communicate with an endpoint then you are using the same or compatible rmw implementations, which includes the serialization technology. The purpose of this section and ROS Service is to provide the information, not to ensure communication can happen.


Agreed. That said, I think this part of the discussion should be preserved, possibly by moving this paragraph out of the TODO section and doing some editing.

… design discussion (#7)

ros-discourse · 2023-03-20T21:31:38Z

This pull request has been mentioned on ROS Discourse. There might be relevant details there:

https://discourse.ros.org/t/ros-2-tsc-meeting-minutes-2023-03-16/30432/1

Signed-off-by: Emerson Knapp <emerson.b.knapp@gmail.com>

ros-discourse · 2023-04-24T20:57:55Z

This pull request has been mentioned on ROS Discourse. There might be relevant details there:

https://discourse.ros.org/t/ros-2-tsc-meeting-minutes-4-20-2023/31087/1

clalancette

We went over this, and we are considering splitting this up into a few REPs:

Type Description, Type Hash, and Type Description Service
Run-Time Interface Reflection - Dynamic Types, Dynamic Data, Dynamic .msg, and typesupport
Evolving message types (REP-2011)

* Full pass removing type descriptions/hashes portion for new rep Signed-off-by: Emerson Knapp <emerson.b.knapp@gmail.com> * Starting a second pass cleaning up the edits Signed-off-by: Emerson Knapp <emerson.b.knapp@gmail.com> * Tiny changes Signed-off-by: Emerson Knapp <emerson.b.knapp@gmail.com> * More tinies Signed-off-by: Emerson Knapp <emerson.b.knapp@gmail.com> --------- Signed-off-by: Emerson Knapp <emerson.b.knapp@gmail.com>

wjwwood · 2023-09-12T21:03:41Z

As of wjwwood#9 being merged, a big chunk of this REP was moved officially into REP-2016 (#381).

wjwwood and others added 8 commits February 15, 2022 15:55

WIP

df394d9

Signed-off-by: William Woodall <william@osrfoundation.org>

still working

c355ca6

Signed-off-by: William Woodall <william@osrfoundation.org>

added more about the enforcing of the type versions

05ac47e

Signed-off-by: William Woodall <william@osrfoundation.org>

Increase readability, fix typos, and reduce ambiguities

6dcb876

Signed-off-by: methylDragon <methylDragon@gmail.com>

Add clarification for field semantics

42cf0de

Signed-off-by: methylDragon <methylDragon@gmail.com>

Implement more fixups

d3449f0

Signed-off-by: methylDragon <methylDragon@gmail.com> Co-authored-by: William Woodall <william+github@osrfoundation.org>

rename from REP-2010 to REP-2011

d7628e7

Signed-off-by: William Woodall <william@osrfoundation.org>

wjwwood self-assigned this Jun 10, 2022

clalancette mentioned this pull request Jun 27, 2022

Backward compatability when adding fields to an existing msg ros2/rmw_fastrtps#620

Closed

methylDragon and others added 9 commits July 7, 2022 17:48

Add alternatives and on_inconsistent_topic section (#5)

b6a7086

Signed-off-by: youliang <tan_you_liang@hotmail.com>

fix rst errors, revise a few sections

a030033

Signed-off-by: William Woodall <william@osrfoundation.org>

touch ups, rename to Run-Time Interface Reflection, reorganize

04306e4

Signed-off-by: William Woodall <william@osrfoundation.org>

diagram typo

d5e27d5

Signed-off-by: William Woodall <william@osrfoundation.org>

fixup

fbc8810

Signed-off-by: William Woodall <william@osrfoundation.org>

change title to a better wording

01edda8

Signed-off-by: William Woodall <william@osrfoundation.org>

fixups, reordering, tools section

9ea2a20

Signed-off-by: William Woodall <william@osrfoundation.org>

wjwwood marked this pull request as ready for review August 22, 2022 05:32

vmayoral reviewed Aug 24, 2022

View reviewed changes

james-rms reviewed Aug 29, 2022

View reviewed changes

wjwwood added 2 commits September 6, 2022 14:00

add feature progress section

24492e5

Signed-off-by: William Woodall <william@osrfoundation.org>

cleanup feature progress list

bd5cdbc

Signed-off-by: William Woodall <william@osrfoundation.org>

This was referenced Sep 20, 2022

Fix complex member printing for DynamicDataHelper eProsima/Fast-DDS#2957

Merged

Sequences and Arrays of Nested Structs Cause Segfault on Pub-Sub for Dynamic Types (with minimal examples) eProsima/Fast-DDS#2969

Closed

methylDragon mentioned this pull request Sep 27, 2022

Dynamic Type Distribution Doesn't Work For Nested Structs (with minimal example) eProsima/Fast-DDS#2984

Closed

1 task

wjwwood mentioned this pull request Oct 11, 2022

new package and interfaces for describing other types ros2/rcl_interfaces#146

Merged

methylDragon mentioned this pull request Oct 11, 2022

Question on DynamicData xtype eclipse-cyclonedds/cyclonedds#1416

Open

This was referenced Oct 12, 2022

.bag2 files? ros2/rosbag2#819

Closed

Type Description Distribution (REP-2011 subset / REP-2016) - formerly "message definitions on-the-wire" ros2/ros2#1159

Closed

clalancette mentioned this pull request Dec 20, 2022

Tracking bug for inconsistent_topic callbacks ros2/ros2#1361

Closed

fujitatomoya reviewed Jan 7, 2023

View reviewed changes

emersonknapp reviewed Jan 10, 2023

View reviewed changes

This was referenced Jan 11, 2023

Update Type Version Hash to use SHA-256, and details updates based on design discussion wjwwood/rep#7

Merged

Calculate type hash from TypeDescription (rep2011) ros2/rcl#1027

Merged

emersonknapp mentioned this pull request Jan 24, 2023

Type hash in interface codegen (rep2011) ros2/rosidl#722

Merged

methylDragon mentioned this pull request Jan 30, 2023

Dynamic Subscription (REP-2011 subset) ros2/ros2#1374

Open

31 tasks

clalancette requested changes Jan 30, 2023

View reviewed changes

Update Type Version Hash to use SHA-256, and details updates based on…

ae78cf9

… design discussion (#7)

methylDragon mentioned this pull request Mar 15, 2023

Add GetTypeDescription.srv (rep2011) ros2/rcl_interfaces#153

Merged

Update RIHS notes based on latest discussion (#8)

ba2f96c

Signed-off-by: Emerson Knapp <emerson.b.knapp@gmail.com>

clalancette reviewed Jun 6, 2023

View reviewed changes

emersonknapp mentioned this pull request Jun 13, 2023

[REP-2016] ROS 2 Interface Type Description #381

Open

MichaelOrlov mentioned this pull request Aug 22, 2023

Plan to port bag migration from ROS 1 ros2/rosbag2#561

Closed

MichaelOrlov mentioned this pull request Sep 17, 2024

Are typesupport libraries needed to be present when deserializing a message in the bag ros2/rosbag2#1809

Closed

	This REP proposes patterns and approaches for evolving message, service, and action ROS interface types over time.
	This REP proposes patterns and approaches for evolving message, service, and action ROS interface types over time. Such capability can be used for evolving ROS 2 itself over time while addressing interoperability issues due to mistmaching type definitions.


		sensor_msgs/msg/Image__RIHS1_XXXXXXXXXXXXXXXXXXXX

		This has the benefit of "just working" for most middlewares which at least match based on the name of the type, and it is simple, requiring no further custom hooks into the middleware's discovery or matchmaking process.


		.. TODO::

		discuss the implications for large messages and the possibility of having the transfer functions be colocated with either the publisher or subscription more directly than with component nodes and remapping.


		These features are described in the following sections.

		Type Version Enforcement


		.. TODO::

		Related TODO, should we just use the TypeDescription described in a later section?


		.. TODO::

		Remove the idea of the user-defined interface version or define how it can be supplied by the user in one or more of the idl file kinds.


		.. TODO::

		It's not clear how we will do this just now, since existing "QoS events" lack a way to communicate this information, I (wjwwood) think.


		.. TODO::

		How can we detect when a remote node is not offering this service?


		.. TODO::

		Should each endpoint be asked about their type/version pair or should we assume that the type/version pair guarantees a unique type description and therefore reuse past queries?


		.. TODO::

		(wjwwood) I propose we use key-value string pairs for the extra information, but I am hoping for discussion on this point.


		.. TODO::

		(wjwwood) A thought just occurred to me, which is that we could maybe have "lazy" Service Servers, which watch for any Service Clients to come up on their Service name and when they see it, they could start a Service Server.


		This section needs to be updated to be inclusive to at least Services, and then we can also mention Actions, though they are a combination of Topics and Services and so don't need special support probably.

		Plugin Interface

[REP-2011] Evolving Message Types #358

Are you sure you want to change the base?

[REP-2011] Evolving Message Types #358

Conversation

wjwwood commented Jun 10, 2022 • edited Loading

Status

Known Issues

ros-discourse commented Aug 22, 2022

james-rms commented Aug 23, 2022

wjwwood commented Aug 24, 2022 • edited Loading

vmayoral left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

amacneil Sep 13, 2022 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

ros-discourse commented Sep 20, 2022

ros-discourse commented Nov 9, 2022

Choose a reason for hiding this comment

Choose a reason for hiding this comment

emersonknapp left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

clalancette left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

ros-discourse commented Mar 20, 2023

ros-discourse commented Apr 24, 2023

clalancette left a comment

Choose a reason for hiding this comment

wjwwood commented Sep 12, 2023

wjwwood commented Jun 10, 2022 •

edited

Loading

wjwwood commented Aug 24, 2022 •

edited

Loading

amacneil Sep 13, 2022 •

edited

Loading