Skip navigation.

exploreopera

| Help

Sign up | Help

more or less

adventures of an itinerant relativist…

Does it Blend? Combining the Best of Tab Activation Modes…

, , ,

Some of the small details can often matter the most. Take this simple conundrum:

Which remaining tab should I activate when I close my current tab?


A simple question indeed, but the answer really depends on what the user expects and he context of what you are doing. Opera have traditionally taken your previous viewed history into account. When you close a tab, you are taken back to the last tab you were looking at. Seems simple and clear.

But other browsers do something different. They take you to the next tab to the right along on your tab bar. In general this may seems a step back, but it has its uses. If you've just opened 5 new tabs from my parent page, moving to the right allows you to read each of the child tabs in turn. Close, move to the new page, read, repeat. And as many users are used to this behavior from their previous browsers, they are confused when Opera takes their viewing history into account instead.

So, trying to cater to this conundrum, Opera have just added some preferences in the latest 9.5 snapshot:

  1. MODE 1: "Activate the last active" — Opera standard mode described above.
  2. MODE 2: "Activate the tab to the right" — The way other browsers do it.
  3. MODE 3: "Activate the first tab opened from closing tab"

The first two options are exactly as described above. But being Opera they've created a new, and cool hybrid mode. The name is pretty confusing, but the behavior isn't. Lets detail how each mode will deal with a specific scenario:

I have 3 tabs, "Page A", "Page B", "Page C". I was reading "Page A". "Page B" has a link I'm interested in and I go to "Page B" and open the link in the background, called "Page D":

In this scenario, "Page A" is the last active tab, "Page B" is the parent of "Page D" and "Page C" is the tab to the right of the current viewed tab. If I close "Page B" using MODE 1:"Activate the last tab", I get to "Page A":

If I use MODE 2:"Activate the tab to the right" I will get to "Page C":

(note: I used the default settings, but if you set "open new tab next to active", then the child "Page D" would have been next to parent "Page B" and MODE 2 would be more useful in this case. Therefore if you want to use MODE 2, I suggest you make sure "open new tab next to active"=ON)

MODE 3:"Activate the first tab opened from closing tab" does the following:


Notice that as "Page B" is the parent of "Page D", then "Page D" is the activated tab (even though it was not physically next to its parent). However if the page I wanted to close has no children, then MODE 3 will use the last active page (in this case "Page A"). Basically this takes the best features of both MODE 1 and MODE 2; it takes your last viewed history into account while respecting parent->child relationships. These three modes allow those comfortable in their ways (MODE 1 and MODE 2 users) to keep their expectations of where they'll end up, but allows power users to use the best of both worlds, a new and elegant mode that adapts to our useage patterns. I just wish it had a more intuitive sounding name!

∃ numeric niceties ∀²

, , , ...

In early 2006, a My Opera user created a highly impressive user javascript to enable Opera to support a core range of MathML. His work had not gone unnoticed by Opera, and it appears at some point Opera hired him. In just over a year he has gone from smart browser user to working on the W3C working group for Opera implementing a standards-based CSS mechanism for MathML rendering:

http://www.w3.org/TR/mathml-for-css/

And that is actually implemented in Kestrel:

http://dev.opera.com/articles/view/can-kestrels-do-math-mathml-support-in/

The benefits of this method is that it uses already existing standards which modern browsers are all working towards; a browser vendor simply needs to insure their CSS support is excellent and have some way to trigger local CSS AFAIK. Mathematicians: exponentially rejoice!


{trivia: I'm no mathematician, but the naming conventions for ∃ means "There exists" and ∀ means "for all" — so my title really reads "There exists numeric niceties for all(squared)}

Access Points for Everything: Part I…

, , , ...

There is a newly created My Opera group (Opera Blue Sky) whose aim is to poke and prod at the very future of web browsing. How do we want to be navigating the web in a few years? The current browser user interface is tired and old. They all have back/forward buttons. They now all have little squares which switch pages (tabs). They all use folders to manage pages you want to remember. There are some differences (which is why some of us use minority browsers like Opera), but the main metaphor is largely unchanged.

Anyway, this is really just a pointer to a little poke at the future that we've been debating for many years in the Opera community, letting access points take over the world browser. Apple have initiated a patent on navigating your browsing history using what are really just access points (remember how M2 can dynamically make relelvant entries into your mail mountain). Several of us Opera users have been suggesting to expand the access point to other parts of the browser UI for years. The original post on Opera Blue Sky is here (also quoted below):

Read more here…

With Opera 7 came a great revolution hidden in an obscure place: dynamic search available through things called "access points" in M2. Once we brethren saw it, we knew it was good. The long-overdue assault on the folder had begun, and as in any war, there was (and still is) much resistance to change. Static folders are comforting, they do what they are told. I put A into B and it stays there. Unfortunately I can't put it into C without copying it, then if I want to edit it I have to do so in two places repetitively and so on. Folders are inefficient and inflexible. They are inferior. Access points can behave like folders, but do much much more. But to benefit from an access point, we need sufficient information to allow the access point the ability to store what we want.

So if we only have a title to a page, that says "Apples are great", we don't know from the title alone if that page belongs to a food category or a computer category. But if we can index the page content then that ambiguity is gone (hint: Kestrel does that now).

So when I saw this patent:

Apple Patent Hints at Future Navigational Interface

I saw there a technology I've been using for years applied to page history; access points. Sorting things in piles dynamically based on categories is what access points do. Indeed, I can do that manually now in Kestrel by using opera:historysearch?q=MY_TOPIC saved as a bookmark, but I want an interface to store my access points more elegantly than that. And I want a "Top 10" for word frequency for my page history. That makes Apple's patent trivial.

Dredging up the Past

, , , ...

So, when I made my five wishes (¿Fem benspænd to Opera developers?), it appears my first wish/obstruction was already granted. When surfing the internet an elephantine memory can be yours, in that Opera indexes your pages and allows that index to be queried later on. The elephantine oracle can be accessed through the address field, and results mix with history items and bookmarks cohesively. Alternatively the oracle is interfaced through opera:historysearch, a search-engine type page. As Opera also has a neat ability to easily make custom search shortcuts, Odegard found you can create a search keyword for opera:historysearch too. Those are the sort of neat little touches that come from having a browser whose features are integrated, not bolted on…

Watch a demonstration…


Opera 9.5: Deep into the Shadowy Underworld…

, , , ...

Well, intrepid alpha adventurers, you probably know that Kestrel has support for text-shadow. But when Opera added it, they didn't just reach parity with the other adventurous browsers that have done so (i.e. Webkit, Konqueror, iCab). Oh no, this is a specification chest-beating contest, so Opera chest-beat even more:

http://nontroppo.org/test/opera/shadows.html

Yep, that psychedelic menagerie, that blinding plurality, that iridescent multiplicity, comes from but one paragraph with six shadows! Six is the limit Opera seems to have chosen to stop infinitely recursive technicolour vomit from invading the internets.

Note however, that the W3C have stated that multiple shadows may be the first thing to cut out of the current CSS3 text effects module. So adventurers, lets make sure we paste techicolour-vomit all over the internets so they'll have no choice but to include it. :devil:

…last to the table is always the cook…

, , ,

Well, as everyone else has already posted, I ran Kestrel through its paces performance-wise. For the couple of hours it took, I tried to be as thorough as possible. That is, unlike the vast majority of performance tests, I give you measures that include the variability of the values I present. This allows you to assess how clearly different two values may be. If I was to be statistically obsessive, I would have had to run several hundred samples for each test for each browser. But life is short, and small-sample studies are published every day in scientific literature world-wide ;-)

Kestrel is pretty phenomenal on the stress tests, and a little bird has told me to expect things are going to get even faster soon!!!

http://nontroppo.org/timer/kestrel_tests/

I've had very mixed responses to the white-on-black design (note images are transparent PNGS — it was an experiment on a small time-budget). The graphs were also a side-project to see how Apple's new spreadsheet app works. Conclusion: crippled for scientists and far more buggy than kestrel!

the curses of hope…

, , , ...

Some sort of collective delirium of hope is spreading like hellfire through the Opera community, a hope which is nevertheless constrained in numerical limitations of 2.5 * 2

Here are my numerically constrained delirious wishes. Most will be somewhat out-of-the-box. I think browsers are stuck in pretty archaic interface paradigms and need a damn good kick in the butt!

  1. Page content indexing — this is simple and a prerequisite for wishes 2 & 3, when you visit a page and it gets added to your history, index the textual content of that page. Visited a page, remember it talked about elasticated lederhosen, but can't find it in your browsers history? Opera's index will get it to you in a flash. This will provide lots of rich meta-data for delirious wish 2:
  2. Unified meta-data interface — many users have long asked that the wonderful dynamic search fluidity of Opera Mail could be applied to bookmarks. I want to take it much further. Opera stores gobs of data all over the place; notes, bookmarks (and their comments, nicks), mail, history (and indexed page content above), tab sessions, current pages etc. All of it is dispersed in its own store, each has a separate search function and UI domain. I want all this information to be indexed and searchable in a unified manner. Bookmarks, notes, mails, web pages in history can be tagged and I can search across domains to get at my data. It will simplify the UI and greatly amplify the way we can link and unify the information we store. It is the modern metaphor of the web: semantic and remixable, applied to the browser interface.
  3. Page Management — we are still stuck with ineffective old metaphors for handling lots of pages; a tab bar which shrinks until it looks like pixel vomit, or spills into anonymous menus. Opera makes it worse by having Mail / RSS / IRC /settings / cache / history pages to add to the mix of web pages; what a goddamn mess! Tabs contain information, they mean or represent something (more so when their textual content has been indexed). A user should be able to organise this chaos. Dynamic groups are the foundation for this; and there are many interesting avenues to explore…
  4. Re-architected Action System — doesn't sound very sexy I grant you. Basically Opera has an internal command system which is hooked into by menu's, gestures and buttons; it is the bedrock on which most of the UI is built!. The current system has been with us since V7 — it is old and creaky. That doesn't stop lots of people making really cool mini-extensions with it (check here and the links contained within), but a *lot* more could be done. This is a fundamental plumbing issue which would allow much more elaborate extensions to Opera's abilities. Ideally a scripting language (ECMAScript engine is already built-in) should replace the logic-challenged old syntax. This gets you many more extension-a-likes without the problems of 3rd-party developers twisting the innards of your browser inside out. Opera need to make a proper functional repository for these customisations (Opera's current system sucks big time).
  5. A new icon; please, pretty please?! Now I use a Mac, the Opera icon is even more revulsive than it was before :yikes:


And what others have deliriously blabbered: UserJS manager, Web developer tools (javascript debugger #1), Auto-deletion of RSS messages after X days and so on and so forth.

Lets get the tags out of the way: ResearchWizard, scipio, Neeraj, robodesign, Eddie, Tomu, operafan2006, Steve (oops, multiplication function has gone awry already).

EDIT: now here is a blog tag we'd love to see: Jon!!!