[e-lang] Returning guards from __getAllegedType: opinions needed

[Mark S. Miller]

Kevin Reid wrote:
> 1. When __getAllegedType is invoked, evaluate all the guard  
> expressions except for those which depend on the results of pattern  
> matches.
> [...]
> 3a. As #1, but evaluate only expressions which are proven to have no  
> side effects.
> [...]
> This would be promoting DeepFrozen to Kernel-E evaluator behavior  
> (but then, so would the planned 'all guards are DeepFrozen' rule).
> [...]
> 3b. Same as #3a, but evaluate the expressions when the ObjectExpr is  
> evaluated. This would avoid repeated work by __getAllegedType and  
> make its output constant, but slow down object creation (outside of  
> the Sufficiently Smart Compiler).

Is there any observable difference between #3a and #3b? If not, then, as far 
as the spec is concerned, this is one option. The remaining differences are 
choices for the implementor.

If this is true, then, I propose that align this decision to the decision 
about whether Kernel-E insists that all guards are DeepFrozen. I propose that 
Kernel-E 0.9 does not require that guards be DeepFrozen, and therefore also 
makes decision #1. However, we also reserve the right to decide that Kernel-E 
1.0 does require guards to be DeepFrozen, and therefore also makes decision 
#3. Therefore, we recommend that code written to E 0.9 not depend on guards 
not being DeepFrozen, and not depend on #1 rather than #3.

-- 
Text by me above is hereby placed in the public domain

     Cheers,
     --MarkM