Group tag block is only partially read since 0bba7e1

[imsodin]

The properties on CommentBlockLine besides the parameterMap (e.g. groupId) aren't populated anymore since commit "Correct parsing of the messages in comment block parameters that contains G and g" / 0bba7e1. I discovered this when upgrading in a unit test that checked those values on example nmea lines.

I don't understand that commit. The commit message claims it affects lines containing g and G, but it actually acts on comments that start with \g:. And the new code seems to be an alernative implementation of the same thing (extract tag keys and values, old code by iteration, new code using String.split), with the difference that it skips handling g, G specially, checksum handling and group-id.

From my understanding the old code was working correctly, however I assume something was fixed by that commit - could you please explain what that is.

[dimitrispallas]

We had a problem in our system parsing the messages with comment blocks that start with the \g: prefix. With that commit our problems has been resolved. Please feel free to apply a pull request in order to completely fix the issue! Thanks in advance!

[imsodin]

Happy to contribute that, however I need to know what exactly was broken for you (I don't see anything obviously wrong with messages starting with \g:) - could you provide some test data that failed and what you'd expect from it (or a unit test)?

[imsodin]

I just checked and the unit test that failed with the new version on my side is also a comment block starting with \g:, and it was working fine before. So I'll definitely need some example to reproduce your issue.

[dimitrispallas]

this is an actual message 24 with comment block that starts with \g:

\g:1-1-21,s:47450,c:1699113260,i:SXSPA:1699115144 I:1699115146 D:1699115146 L:FM136 G:hndgs T:0 1261!AIVDM,1,1,,A,H7Oks<0hTtpE=<0000000000000,24C

Beforehand our system wasn't able to decode and produce the JSON output for that kind of sentence. And we are parsing millions of them in a daily basis. As I already said currently everything works properly in comparison with the previous code.

[imsodin]

I got that, I just need to figure out what's wrong to produce a change that works in both cases - thanks for providing an example for that :)

I'll probably have a look at this next week.

[imsodin]

Obviously that "next week" prediction didn't pan out xD.
It's basically in my backlog now, but given no-one else seems to be affected it anyway is hardly time-critical.