On 2008-05-19, at 14:42 EDT, Donald Anderson wrote:
On May 19, 2008, at 2:19 PM, P T Withington wrote:
2) I wonder if you would get around your space explosion if you
interned the file name strings? That would also speed up comparing
them.
I don't think that helps. We are embedding the strings within
longer and longer strings
as the unparser is making the entire program bottom up.
So:
"i = i + 2;"
might become
"\u0001Finputfile.lzx#234.5\u0001 i = i + 2;"
I was confused about where these annotations were going. I didn't
realize they were going into the 'giant unparser string'.
I guess we should have an improvement filed to make the unparser
smarter. Or to store annotations in a side table by offset, or
something. But not for now.
