Hi Carlos,
This is definitely out of my area of expertise. I spent some time poking
through the code.
First, regarding your question about whether the parameter should be
AMFBInaryData or BinaryData, IMO, the answer should be that the parameter
should be an interface. If there aren't backward compatibility constraints, we
should try to use interfaces more in new APIs. In looking around, I noticed
that there is:
Core/src/main/royale/org/apache/royale/utils/IBinaryDataInput.as
Network/src/main/royale/org/apache/royale/net/utils/IDataInput.as
Storage/src/main/royale/org/apache/royale/storage/file/IDataInput.as
Seems like there should either be only one, and AMFBinaryData and BinaryData
should both implement it, or maybe one interface should extend the other. So
some refactoring is probably warranted there.
Now to your main question. Are you sure that code 168 is an AMF type code? I
could certainly be wrong, but it occurred to me that the 168 could just be the
first byte in serialized data of the message and that the AMF decoding should
instantiate the object based on the class alias then see if it implements
IExternalizable and call readExternal without bothering to examine that byte in
the stream.
HTH,
-Alex
On 5/31/18, 4:57 PM, "carlos.rov...@gmail.com on behalf of Carlos Rovira"
<carlos.rov...@gmail.com on behalf of carlosrov...@apache.org> wrote:
To be more clear on this, what I need is some guidance on how to deal I
readValueObject(), EXTERNALIZED_OBJECT case and if I must pass the rest of
data to the AcknowledgeMessageExt.readExternal method, or pass the entire
AMFBinaryData to that method. Or another thing I don't have in mind.
thanks!
2018-06-01 1:37 GMT+02:00 Carlos Rovira <carlosrov...@apache.org>:
> Hi,
>
> in order to finish Small Messages implementation I need some guidance.
> Hope you guys could throw some light on this.
>
> When turning small messages on, BlazeDS send serialized version of some
> messages. For example the most important is "*DSK*" (the
> registerClassAlias for *AcknowledgeMessageExt.as*), in this case we get a
> type code *168* in "*readObject*" method from *AMFBinaryData*. We can get
> as well *169* and *161* codes, don't know if there's more since I
> couldn't find any documentation about this.
>
> So in *readValueObject*() I introduced a new case *EXTERNALIZED_OBJECT*
> (that is a constant with 168 value)
>
> Inside we need to deal with the deserialization of the
> *AcknowledgeMessageExt* message (the *DSK*).
>
> This kind of messages like the rest implements all *IExternalizable*
> methods (*readExternal*, *writeExternal*), that receive a *IDataInput*
> and *IDataOutput* respectively.
> I get all this code of messages from Flex RPC compile correctly . Here I
> have a doubt:
>
> ** When I have a ByteArray param here it should be BinaryData or
> AMFBinaryData in Royale?* (I suppose the second, but can't ensure it)
>
> Finally the main problem...How can I deserialize the DSK object that comes
> vía AMF
>
> 1.-* I'm pretty sure that AMFBinaryData must implement IDataInput and
> IDataOutput since is the object I suppose I must to pass to readExternal
or
> writeExternal right?*
>
> 2.- In *readValueObject*(), case *EXTERNALIZED_OBJECT* I tried to do
> something like:
>
> *var dsk: AcknowledgeMessageExt = new AcknowledgeMessageExt()*
> *dsk.readExternal (this);*
> *value = dsk;*
>
> this is not working, if not we'll be done ;)
>
> But as I don't know how flash IExternalizable works under the hood, I
> don't know how to deal with this in Royale.
>
> I suppose that Flash Player manages the readExternal and writeExternal
> methods under the hood, right? but in Royale I suppose we need to call it
> ourselves.
>
> Maybe the IExternalizable, IDataOutput and IDataInput are not right as I
> write those. Notice that the last two doesn't have all the methods in
> Flash. I write only the required ones. I'll write the rest as we get the
> current problem solved.
>
> As I said, all messages has the message methods for readExternal and
> writeExternal compile in royale, so I think we are so close to get this
> implementation working hopefully with some help for your part to finally
> get the final pieces of this puzzle solved :)
>
> Thanks in advance
>
> Carlos
>
>
>
>
>
> --
> Carlos Rovira
>
https://na01.safelinks.protection.outlook.com/?url=http%3A%2F%2Fabout.me%2Fcarlosrovira&data=02%7C01%7Caharui%40adobe.com%7Ceffcd4e8bcf14b3d51dc08d5c75238cc%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C636634078327400307&sdata=iaUSnW4FfeZFtb2q0Hd4aHTUW5wgNyzbRIYfna7WD9M%3D&reserved=0
>
>
--
Carlos Rovira
https://na01.safelinks.protection.outlook.com/?url=http%3A%2F%2Fabout.me%2Fcarlosrovira&data=02%7C01%7Caharui%40adobe.com%7Ceffcd4e8bcf14b3d51dc08d5c75238cc%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C636634078327400307&sdata=iaUSnW4FfeZFtb2q0Hd4aHTUW5wgNyzbRIYfna7WD9M%3D&reserved=0
