Re: [PATCH 3/4] dfa: refactor to prepare for upcoming optimizations

[Paolo Bonzini]

On 06/07/2011 01:34 PM, Jim Meyering wrote:
> It's slightly tighter and more readable to make this temporary "const".
> Only down-side is that inserting "const " pushes line-length to 80;-)
> 
> ACK either way.

Since the const would disappear on the next patch, I pushed it without.

On the other hand, before pushing I changed the else branch from this:

+        addtok (CSET + work_mbc->cset);

to this:

+        {
+          /* The single-byte character set must be non-empty, or due to the
+             test above the entire MBCSET would be empty (which is invalid).  
*/
+          assert (using_utf8() && work_mbc->cset != -1);
+          addtok (CSET + work_mbc->cset);
+        }

It all goes away in the next patch too, but the way that patch changes the
code is clearer with the assertion in place.

Paolo