<br><br><div class="gmail_quote">On Tue, Jun 15, 2010 at 3:53 PM, Andreas Raab <span dir="ltr"><<a href="mailto:andreas.raab@gmx.de">andreas.raab@gmx.de</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;">
<div class="im">On 6/15/2010 3:32 PM, Eliot Miranda wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
Clearly with the 20/20 scheme its easy to steal a couple of bits for<br>
more GC flags by taking a bit from each 20 bit field. Further, with<br>
Squeak, the inst size byte might be pointless (although contexts are<br>
still indexable objects with fixed fields) and so I doubt bits are at a<br>
premium. Hence the real question is whether I should be whorish and go<br>
for the 16-bit compact fast class index access or not.<br>
</blockquote>
<br></div>
My $.02: Start with 16 bits and declare the free bits "reserved". When someone complains for the first time, review and decide how many more bits to devote. Rationale #1: Rather than taking away resources when you notice you need an extra header bit, start with the smaller number but keep some room for growth if it becomes necessary. Rationale #2: YAGNI (i.e., we really don't know if we'll need more header bits or if we need more room for classes).<br>
</blockquote><div><br></div><div>Excellent suggestion. </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;">
<br>
Cheers,<br><font color="#888888">
- Andreas<br>
<br>
</font></blockquote></div><br>