Hi, On 8/31/07, Emmanuel <[EMAIL PROTECTED]> wrote: > Hi Dogacan, > > I'm wondering if you have done some works on this subject. Could you confirm > ? > Do you want me to work on it and provide a patch ?
I haven't worked on it, so feel free to work on it and open a JIRA issue. Btw, I was wrong in my earlier post (We don't re-order normalizers again and again), but still, not instantiating a new object can be a big win. > > > > On 8/2/07, Emmanuel <[EMAIL PROTECTED]> wrote: > >> I've got a simple question why do we normalize each single outlink int > >> he > >> constructor of the object. It involved the creation of many > >> URLNormalizer > >> object. > >> > >> We could just add the normalizer in ParseOutputFormat just before the > >> filter > >> and it will limited the number of instanciation. > >> Don't u think ? or did i miss something ? > >> > > > > > > I am not sure, but I think the idea is to make Outlink class useful > > outside of ParseOutputformat (so that if you use Outlink w/o > > ParseOutputFormat, you would still end up with a normalized url). > > > > However, this minor advantage is hugely offset by the fact that we are > > recreating URLNormalizers for every outlink (and if you have an > > ordering on your normalizers, re-ordering them *every* *single* time), > > so overall moving normalizing into ParseOutputFormat seems like a good > > idea to me. (and while we are doing that, perhaps we can stop creating > > a ParseUtil instance for every ParseSegment.map [even though it has a > > smaller overhead]). > > > > -- > > DoÄŸacan Güney > > > -- Doğacan Güney
