On 10/19/2011 08:31 PM, Hai Dong Li wrote:
This email is just for your attention. I'm relatively new to work
in a
community, so I didn't pay much attention to the readability of the
comments last email. It seems comments lie in a large patch like this is
easily to be omitted. So I cut the codes, leave codes associated with
the comments.
Yes, trimming to just relevant context is a must for any high-volume
patch list. Also, separate your replies from the quoted material by
blank lines, so it stands out better (visually, I find it easier to spot
replies that appear in isolation, by scanning just the left column; not
to mention that some mailers corrupt long lines on quoted replies where
a long single-line paragraph in the original turns into a wrapped
multi-line text with the first line quoted but subsequent lines
unquoted; adding whitespace before your reply makes it obvious that you
made the comment, rather than your mailer reformatting things).
> + virBufferAdjustIndent(buf, -2);
> + if (virBufferGetIndent(buf, false) != 1 ||
> + virBufferGetIndent(buf, true) != 1 ||
> + virBufferError(buf)) {
> + TEST_ERROR("Wrong indentation");
> + ret = -1;
> + }
So now buf->indent is 1. Go to the next step, the indent is given -2
again, see what will happen.
if virBufferAdjustIndent failed to check the indent overflow, the
buf->indent will be -1,too, so it may avoid the check
(virBufferGetIndent(buf, false) != -1) and (virBufferGetIndent(buf,
true) != -1).
> + virBufferAdjustIndent(buf, -2);
So I think -3 may be better.
Good idea; I've folded that into my patch.
--
Eric Blake eblake(a)redhat.com +1-801-349-2682
Libvirt virtualization library
http://libvirt.org