2 Mar
2017
2 Mar
'17
1:35 p.m.
On 02/03/17 14:22, David Matthews wrote:
Adding support for interpretation is going to be a bit of work and until it's done and tested there's no way of being sure it will solve this problem. I'd prefer not to delay the release.
Interpretation will not really solve the "by eval" problem: code generation produces huge ML modules and expects that they run efficiently. I've proposed that merely as a measure of last resort.
Since you say that code GC should basically work (again), this is what really counts.
Makarius