On Fri, 2011-12-09 at 12:34 -0700, Eric Blake wrote:
> On 12/08/2011 05:38 PM, Eric Blake wrote:
> >> int i;
> >> for (i = 0 ; i < strlen(model) ; i++) {
>
> Hmm - an O(n) function call on an O(n) loop, for a quadratic action (of
> course, it's in the noise, since the user's model name is likely short).
> But we can do better with a more efficient search for bogus bytes
> (strspn is O(n), if implemented well in libc).
>
> >> - int char_ok = c_isalnum(model[i]) || model[i] == '_';
> >> - if (!char_ok) {
> >> + if (!c_isalnum(model[i]) && model[i] != '_' && model[i] != '-') {
> >
> > I'm not sure if we need to tweak our RNG grammar to also allow this in
> > the XML validation. I'll check that out tomorrow, when I get around to
> > applying this one and reviewing the rest of the series.
>
> It turns out the XML didn't do any validation at all. Here's what I
> came up with - tightening the RNG and relaxing the domain_conf code, so
> that they now match. Since the concept is the same as yours, I went
> ahead and pushed it, but I claimed authorship on this one, since I
> practically rewrote it.
Yep fair enough. Your new version is much nicer, looks good.
cheers
Attachment:
signature.asc
Description: This is a digitally signed message part
-- libvir-list mailing list libvir-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/libvir-list