[C++] Unclear documentation for adding schemas

[jarkenau]

The documentation for C++ includes a minimal example on how to add schemas. For writing a simple std_msgs/String it can be done like this:

...
// Register a Schema
mcap::Schema stdMsgsString("std_msgs/String", "ros1msg", "string data");
writer.addSchema(stdMsgsString);

// Register a Channel
mcap::Channel chatterPublisher("/chatter", "ros1", stdMsgsString.id);
writer.addChannel(chatterPublisher);
...

Where string data describes the layout of the message. How should the string look like for more complex & nested messages, e.g sensor_msgs/Image?

[foxhubber[bot]]

Linear: FG-1855

[jtbandes]

The format for ROS schemas is documented here: https://mcap.dev/specification/appendix.html

[jarkenau]

Thanks for the hint 👍 . But it still won't work as expected. As described in the docs here I used gendeps --cat to generate the description and saving sensor_msgs/Image now works just fine. If I open the written .mcap file in foxglove studio, the image panel throws the following error Cannot read properties of undefined (reading 'stamp'), am I missing something?

[jtbandes]

Looking at your schema data, it appears the lines are concatenated without any \n newlines, so the schema is treated as one large comment, and thus the app doesn't think it contains any fields.

[jarkenau]

Yes, I forgot to add the \n. But after adding them, I run into another error 😕

RangeError: Offset is outside the bounds of the DataView
    at DataView.getUint32 (<anonymous>)
    at MessageReader_StandardTypeReader.uint32 (file:///opt/Foxglove%20Studio/resources/app.asar/renderer/3639.js:130:2353)
    at new Record.std_msgs_Header (eval at MessageReader_createParsers (file:///opt/Foxglove%20Studio/resources/app.asar/renderer/3639.js:142:39), <anonymous>:17:23)

Again the output file:
macp-file.zip

[jtbandes]

Your messages have no data:

$ mcap cat output-3.mcap 
1676317831042191294 /image [sensor_msgs/Image] []
1676317832049974546 /image [sensor_msgs/Image] []
1676317833074946271 /image [sensor_msgs/Image] []

[jarkenau]

Thanks for the help, I managed to figure everything out. Do you think a note should be added to the docs, to pay attention that the \n are present?

[jtbandes]

The documentation states the files should be concatenated, and you can use gendeps --cat to produce the correct format. What additional clarification would be helpful? You can make a PR to update the docs if you feel there is a useful change to be made, but personally they seem clear enough to me.