Magic_numbers in calculateMaxSerializedSize

[mikaelarguedas]

This is a follow-up of #36
There are still magic numbers (101, 257 and 100) use to determine the maximum size of a sample.
It would be good to investigate if we can get rid of them and improve the logic of the MaxSize calculation

[allenh1]

@mikaelarguedas This might also be something I could work on

[mikaelarguedas]

@allenh1 Is this one still on your radar ?

[allenh1]

Yep -- I was waiting on the refactor to finish, but looks like I can go after it now.

[mikaelarguedas]

There is now only one magic number left.

@richiprosima I see there is a todo in your name to modify it, do you have any insight as of if this variable should be kept ?

[richiprosima]

That value is inherited from when there was only preallocated memory policy in FastRTPS, and dynamic types had to have a maximum length. For sequences/arrays the maximum length was 100. Now ROS2 uses preallocated_with_realloc memory policy and maximum lengths are not needed.

[dirk-thomas]

@richiprosima can you please propose a patch how to get rid of this variable. It isn't clear to us how to remove it cleanly.

[richiware]

I'm working on it

[richiware]

PR #152 removes those magic numbers.

[mikaelarguedas]

All magic numbers have now been removed 🎉 closing