In the next day or two, I will be checking in some changes for Unicode that makes the utf-8 reader much faster (2-3 times or so). The other formats don't see much change in speed. However, to bootstrap this change you will need to do a cross-compile using the unicode-string-buffer-branch. Use boot-2009-10-1-cross.lisp as the cross-compile script. (Don't forget: you need to do, eventually, an x87 and an sse2 build. Non-unicode isn't needed here.) Once you finished the cross-compile, you can use that to build the sources. You can try this with the unicode-string-buffer-impl-branch now, or wait until that branch is merged to the trunk. Ray
Raymond Toy wrote:
In the next day or two, I will be checking in some changes for Unicode that makes the utf-8 reader much faster (2-3 times or so). The other formats don't see much change in speed.
Changes have been merged. One thing I forgot to mention is that you need to use boot-2009-10-01.lisp to compile these sources for a non-unicode build. You can use 20a for that instead of a build from unicode-string-buffer-branch. Ray
participants (2)
-
Raymond Toy -
Raymond Toy