machine learning is needed because a substring search may match a large number of results. In order to "just work" the first result should be the "one you want more" and to do that google chrome need machine learning.
Daniele S. 2009/12/3 rojer <[email protected]> > On Dec 3, 5:17 am, Chen Yong <[email protected]> wrote: > > I believe this would behttp://crbug.com/367 > > sort of. although i don't know what machine learning has to with this, > at all. > give me substring search, that's all i really want. > > ok, i voted for that bug. but no solution in sight, it seems... > > > > > On Thu, Dec 3, 2009 at 3:41 PM, PhistucK <[email protected]> wrote: > > > You can start by filing a issue at crbug.com. See if it does not get a > > > "WontFix" or any rejections here. If they say they are willing to > accept > > > patches for this, you can start working on it. > > > > > Reply with the issue URL here. > > > > > ☆PhistucK > > > > > On Tue, Dec 1, 2009 at 22:16, Rojer <[email protected]> wrote: > > > > >> Hello, fellow Chromium users and devs. > > > > >> This is probably my biggest gripe with Chrome, and i can stand no > > >> longer... > > > > >> History searches by URL from the Omnibar are awful! > > >> If you recall part of a URL, hardly ever does Omnibox return useful > > >> results. > > > > >> Suppose you havehttp://some-site.com/potentially/long/url/ > > > > >> The *only* way to get Chrome to dig up that url seems to be to start > > >> entering "some-site.com/p" etc. > > >> If you happen to recall any other part of the url, searches or > > >> suggestions will bring up nothing. > > >> This is very bad, especially considering how good Firfox is at this in > > >> comparison: it will start suggesting as you type any part of the url > > >> whatsoever. Chrome, on the other hand, seems to be tokenizing URLs on > > >> slashes (not sure about this, though). > > >> if you type "long/url" it will treat "long" as a hostname and > completely > > >> give up on any suggestions. > > >> in other words - really, really bad experience searching parts of > URLs. > > >> give me arbitrary substring searches! ok, maybe not arbitrary > substring. > > >> if Chrome insists on tokenization, fine. but i should be able to find > > >> this history entry by typing e.g. "long url" or "potentially long". > this > > >> doesn't happen at the moment. > > > > >> Can something be done about this? > > >> I am willing to spend some time and contribute code towards fixing > this > > >> unless it'd be overly complex in the current architecture. > > > > >> Thanks > > > > > > -- > Chromium Discussion mailing list: [email protected] > View archives, change email options, or unsubscribe: > http://groups.google.com/group/chromium-discuss > -- Chromium Discussion mailing list: [email protected] View archives, change email options, or unsubscribe: http://groups.google.com/group/chromium-discuss
