You're right, I've never imagined a system that would need to be
flipping the mode back and forth.  In that case, adoptInstance is
quicker, but since we've had to special-case it, its certainly more
"complex" and the fact that swapping out an objects class is not a
common thing to do in Smalltalk, it seems fair to say its less
legible.  That's not an objection, though.  :)

On Tue, May 31, 2016 at 11:12 AM, Levente Uzonyi <leves at caesar.elte.hu> wrote:
> On Tue, 31 May 2016, Chris Muller wrote:
>
>>> It's an optimization, so it's not mandatory. Of course it's nice to have,
>>
>>
>> An optimization of what?  Execution speed?  Is it such an optimization
>> that any human could ever possibly notice?  Setting the mode of a
>> stream, really?
>
>
> I suspect that you expect one to set mode at most once per stream, which is
> not always the case.
> I could easily give you an artificial benchmark showing 30x speedup, but
> that would make no sense.
> It's a library method and as such, why not make it quicker when possible?
>
>>
>> I like most of your optimizations but, IMO, this one is so
>> infinitesmal, I don't think it exceeds the cost of the decreased
>> legibility / complexity..
>
>
> Please explain the legibility / complexity part.
>
> Levente
>
>>
>>
>