IonRawBinaryWriter_1_1 annotations#661
Conversation
| * [writeAnnotations] may be called more than once to build up a list of annotations. | ||
| */ | ||
| fun writeAnnotations(annotation0: Int, annotation1: Int, vararg annotations: Int) | ||
| fun writeAnnotations(annotations: IntArray) |
There was a problem hiding this comment.
🗺️ I realized that it (a) made the implementation cleaner and (b) just made sense to be explicit about the array here, so I changed the function signature.
There was a problem hiding this comment.
Does this ever need to be called from Java? What does that look like?
There was a problem hiding this comment.
For now, I don't think the Ion 1.1 raw writer needs to be called from Java. However, it is callable from Java. IntArray is Kotlin's name for int[].
As for removing the vararg parameter—instead of this:
writer.writeAnnotations(1, 2, 3, 4, 5);Now you have to be explicit about the array:
write.writeAnnotations(new int[]{1, 2, 3, 4, 5})This is a reversible decision.
| var isDelimited: Boolean = false, | ||
| var position: Long = -1, | ||
| var length: Long = -1, | ||
| var length: Long = 0, |
There was a problem hiding this comment.
🗺️ As I was working on annotations, I realized that it made more sense to always initialize the length of the ContainerInfo to 0. I also added a clear() function after I realized I might be leaking state between containers.
|
|
||
| // Update the container length (unless it's Top) | ||
| if (currentContainer.type != Top) currentContainer.length += numBytesWritten | ||
| currentContainer.length += numBytesWritten |
There was a problem hiding this comment.
🗺️ A long is pretty big, and it doesn't hurt anyone to have the length of Top being tracked, so I got rid of the branching here (and similarly in stepOut()).
| * [writeAnnotations] may be called more than once to build up a list of annotations. | ||
| */ | ||
| fun writeAnnotations(annotation0: Int, annotation1: Int, vararg annotations: Int) | ||
| fun writeAnnotations(annotations: IntArray) |
There was a problem hiding this comment.
Does this ever need to be called from Java? What does that look like?
| * are being added one by one. | ||
| */ | ||
| private inline fun ensureAnnotationSpace(n: Int) { | ||
| if (annotationsIdBuffer.size < n || annotationsTextBuffer.size < n) { |
There was a problem hiding this comment.
Should these grow independently? Does that significantly complicate things?
There was a problem hiding this comment.
The way I have it implemented right now, if position x of annotatationsTextBuffer is null, then we know to use the value from annotationsIdBuffer, so it does simplify the write logic if annotatationsTextBuffer is at least as large as annotationsIdBuffer, and if one has to be as least as large as the other, it simplifies the "growing" logic if we just make them the same size. I am not too concerned about the efficiency of growing both arrays at the same time because realistically, we don't expect there to be 10+ annotations for nearly all use cases.
There was a problem hiding this comment.
I wonder if this should be capped at 1 rather than using the same value used for containers. It seems less likely that your annotations would grow larger than can be described in a 1-byte FlexUInt.
That said, making a change here is probably only valuable if it simplifies the code.
There was a problem hiding this comment.
It does marginally simplify the code because the minimum we always need to preallocate is 1, so if we're also capping it at 1, then we can just hard-code it to 1.
There was a problem hiding this comment.
What do you think about changing the spec so that this is 0x90 instead? That way it at least aligns with the upper nibble type (9 = symbol, whereas 7 = timestamp)
There was a problem hiding this comment.
I like that, and I suspect that was probably the intention, but then parts of the spec were modified without fully updating other parts of the spec.
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## ion-11-encoding #661 +/- ##
==================================================
Coverage ? 64.64%
Complexity ? 5703
==================================================
Files ? 199
Lines ? 24695
Branches ? 4386
==================================================
Hits ? 15963
Misses ? 7433
Partials ? 1299 ☔ View full report in Codecov by Sentry. |
2 participants
Issue #, if available:
None
Description of changes:
There are two commits in this PR.
The first commit refactors some tests. It adds an
autoCloseoption to theassertWriterOutputEqualsfunction and anassertWriterThrowsfunction, and it updates the test cases to use those.The second commit adds annotations to
IonRawBinaryWriter_1_1.By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.