[V3dot10] Re: Tentative of start 3.11

[Keith Hodges]

Dear All,
>> You mean, take a 3.8.2 release, load all the bug fixes in 3.9
>> and 3.10 except for Traits and Universes, and sell it as a
>>     
If you consider the amount of work that went into 3.9 I myself doubt
this is a sensible option.
>> simpler release image without Traits? Sounds reasonable. For the
>> forseeable future, we will need to maintain an image without
>> Traits, if only for Croquet and Etoys folk. And figuring out how
>> to load and unload traits will be a lot of work, and no one will
>> want to load or unload traits themselves anyway
>>     
I thought that the whole point of Traits is that they can be flattened.

I think it would be less work and more beneficial to bring the etoys and
croquet work up to 3.10, even if the 3.10 image requires it's traits to
be flattened.

If the issue is binary portability, then my first comment is "how dumb",
not building a binary format that can cope with the future is doomed.
MC1 did just that. My second comment is that there are numerous tricks
that MC1.5 uses to preserve binary portability, so it is doable and you
can even future proof your classes.

If we had a namespace solution, we could have the old class definitions
in a second namespace, load the instances in and migrate them into the
new formats in the main namespace. My process specific global remapping
technique might work for this task.

I sincerely believe that there is more than one way to skin a cat, and
no problems are insurmountable.

So the question for me is , what are the real issues preventing Croquet
and eToys from porting to 3.10/11 etc.

Keith