On Mon, May 2, 2011 at 2:30 PM, Greg Stein <[email protected]> wrote: > On Mon, May 2, 2011 at 13:58, <[email protected]> wrote: >>... >> + if (info2->wc_info->checksum >> + && info2->wc_info->checksum->kind == svn_checksum_md5) >> + info->checksum = svn_checksum_to_cstring( >> + info2->wc_info->checksum, pool); >> + else >> + info->checksum = NULL; > > I would note this means the checksum will be NULL for users of the old > API (since the new API returns SHA-1 checksums). This could be viewed > as a regression.
The docs for the old struct indicate that the field may be NULL, so callers need to be aware of that possibility aforehand. Also, the regression isn't caused by the above change, but rather our decision to change which checksum we return. We've been returning sha1's for weeks; at least now we let people know that fact. :) -Hyrum
