Peter N. Lundblad wrote:
> Hi,
>
> While testing the new diff stuff (ignore whitespace etc.), I changed the
> token read chunk size from 128K to 4, and found an old corner case bug.
> The problem is that the code will return a spurious empty token if the
> file is exactly n*CHUNK_SIZE bytes long (n>0). This confuses the diff
> algorithm, which thinks that the file has an extra line. Attached is a
> fix for this, which seems a little clumsy. If anyone has a better
> solution, feel free to suggest/commit.
The fix makes sense. Though I'd avoid calculating the hash as well.
> Note that this is in 1.3 (and probably earlier)
Yes, this has probably been there since the start. Nice catch.
Sander
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Wed Feb 8 01:34:36 2006