rekonq: working on “suggestions” 2

Just a brief update to previous post. I tried following discussion and “suggestions” there and implemented a sort of “mix” of them. I’m quite satisfied from results. A screenshot will let everything being simpler:
rekonq suggestions for 0.9: second proposal.
So, what happened here:
results are (no more than) 10 now. In 0.8 they are no more than 8. You have all sites matching, but the ones “startingWith” are considered more relevant and are listed before. The other relevance criteria (number and last date of visits) are anyway considered and help listing history items.
Compare this screenshot with actual implementation and with first proposal to have a clear idea about.

I tested it all the morning and it really seems “matching” better my needs. So I’m probably going on submitting a review request for.
Last chance to submit comments about 🙂

12 responses to “rekonq: working on “suggestions” 2

  1. Nice to see work on Rekonq. I would like it to have zooming as in Chromium, with Chromium I can set the zoom level for every page on e.g. 80%.

  2. Nice work ! Thanks to listening to users suggestions 🙂
    It’s not really important but is it possible to hide the line http://kde when it’s not matching a complete or valid url (with .org .com…) ?

    Happy New Year 🙂

    • uhm… no. In fact you can call your pcbox “kde” or “pingpong” or “thenameyoubetterlike” and enable there a web server. So, to remove the browse “http://something” you should ping it to see if it really exists or not.
      The actual rule is just check for something like you suggest (startsWith http, finish with dot com/net/somelse) and eventually show it before or after in the list.

  3. – i think this is a great improvement, good job!
    – replace the word “actual” by “current” when you write in english and I’ll be unable to tell that you’re not a native speaker 🙂
    – i remember running into some issues because there was no sure fire way to filter out search results page from the suggestions (the algorithm depended on which search engine were configured in rekonq, and didn’t work if none was selected). Did you find a solution to filter the search results in a more robust way ?

    • MPomme, this about language is the best compliment I ever received, many thanks for 😀
      About search engine issues, the proposed mechanism “should” cut out this problem, trying to guess user needs instead of letting “google” (or search engine) suggestions match it.
      We’ll see if it really works.

  4. It always bugged me that when I enter “pl” in Chromium it automatically selects “planetkde.org” and I hit enter and go. In rekonq it always trys to search google for … Poland 🙂

    • That’s exactly the problem this mechanism tries to accomplish: not just getting sure the result is in your list, but show it in the first lines (hopefully, first).

  5. Some suggestions for small usability improvements:

    – Exclude entries with invalid URLs (like “http://kde” in the screenshot) from being shown.
    – Print the URLs in a different color (maybe gray?) than the titles, so it’s easier to quickly skim over the titles with your eyes.
    – Maybe show HTTP URLs without the “http://” prefix – it would improve readability, and everyone would still understand that it’s supposed to be a HTTP URL. (PS: Firefox also does this in its Awesome Bar.)

    • I explained some lines before why your first point cannot be satisfied without a performance impact.
      I can easily work on other 2 suggestions to add them to this code. many thanks for.

  6. Yep that http:// part definitly needs to go, hide it anywhere. It just annoys. (Weird that Google Chrome hides it while on the Android where screen space is quite limited it does show it)

  7. Pingback: Rekonq, i suggerimenti nella digitazione sulla barra degli indirizzi | Indipedia – Indipendenti nella rete

  8. +1 vote for the http:// removal. How does Chrome+Firefox solve the problem you have when it’s removed?

Leave a comment