Revisiting issue 17

[William Clocksin]

Although the issue seems to be fixed and closed, it is still happening in the application. Here are the first few lines of the JSON I get when I try to serialise a document. Notice in line 3 (depending where your page wraps), where the hangers field is supposed to serialise to a list where the first element is a calliope.TextBox. Its area field gets serialised to an object called "@id":37, which ought not be there, and is over 400 lines long (not shown). My custom serialiser correctly serialises the area (as shown in the report for Issue 17) but only when the TextBox is not in an ArrayList. When it is in an ArrayList as shown here, it doesn't do the custom serialisation.

{"@id":352,"@type":"calliope.Document","pages":[{"@id":1,"@type":"calliope.Page","document":{"@ref":352},"format":"PGCENTRE","alignment":"NONE","topsys":0,"botsys":0,"topEdge":0.0,"bottomEdge":0.0,"fillheight":698.0,"num":0,"type":"PAGE","bb":null,"hangers":[{"@type":"calliope.TextBox","subtype":7,"xoff":282.0,"yoff":290.0,"area":{"@id":37,"highlightTextFill":{"@type":"org.fxmisc.richtext.CustomStyleableProperty","bean":{"@ref":37},"name":"highlightTextFill","cssMetaData":{"@type":"org.fxmisc.richtext

[William Clocksin]

PS: I downloaded and installed the Master code before testing this.

[John DeRegnaucourt]

Please send a small test example that replicates the problem. If you look at CustomWriterTest.testCustomerPersonWriterReaderinCollectionTypes() and CustomWriterTest. testCustomerPersonWriterReaderForCollectionFields() both of these tests caused the reported issue to show up. A fix was made and these tests now pass. Looking for another test case that proves incorrect behavior, but with very small test objects, so that it can be bundled up into a JUnit test case.

…

On Tue, Oct 31, 2023 at 10:08 AM William Clocksin ***@***.***> wrote: PS: I downloaded and installed the Master code before testing this. — Reply to this email directly, view it on GitHub <#201 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABI76TQXS3I7CJGPXPUERRTYCEA6LAVCNFSM6AAAAAA6XYHSAWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTOOBXGI4TGNJSGE> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[William Clocksin]

Yes, fair request. It may take some time to prepare a small test case, as this is in a program of some 56,000 lines.

[John DeRegnaucourt]

If you look at the People, Person, Pet classes I created for those tests I mentioned - they are simple but have enough structure to repeat the issue. Repeating the issue with classes not specific to any project, that is an important part of getting the repeatable test built. Note, I used static inner classes, which make it easy to package the "data" classes with the test case.

…

On Tue, Oct 31, 2023 at 11:16 AM William Clocksin ***@***.***> wrote: Yes, fair request. It may take some time to prepare a small test case, as this is in a program of some 56,000 lines. — Reply to this email directly, view it on GitHub <#201 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABI76TQN4MZR3JMLXRMCI6TYCEI5FAVCNFSM6AAAAAA6XYHSAWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTOOBXGQZDQOBXGQ> . You are receiving this because you commented.Message ID: ***@***.***>

[William Clocksin]

Understood. One difference between the tests and my code is that my classes are all subclasses two or more deep, where your People and Pet are base classes (not extensions of any class other than Object). I wonder if that matters.

[William Clocksin]

Very embarrassing: My testing code to test your fix was faulty. Your fix indeed was correct. In one of my tests of your fix, I had written
arg.put(JsonWriter.CUSTOM_WRITER_MAP, Serialising.getCustomWriters());
after the
json = JsonWriter.objectToJson(obj, arg)
and not before. This is why I seemed to get the problem in one of my tests, but not all. Mea culpa. Issue closed.