Slashdot Mirror
OmniWeb Announces 5.0 Browser
wcbrown writes "OmniGroup, makers of the popular Mac OS X browser OmniWeb have announced the upcoming beta of their next-generation browser. There's going to be tabs and they're not like any other browser out there. There's going to be a way to save and share your browsing state so you can restore your window locations and the URLs in them. There's going to be some cool nice-to-haves like integrated RSS reading, per-site preferences, and search shortcuts. The beta will be available February 2, 2004."
When I first saw the way Omni had implemented tabs in OW, I thought they were trying to be different for its own sake.
On this thread, Tim2, who's on the team at Omni, explains the reasoning behind their tabs implementation (vertical tabbing, drawer as opposed to hotlist a la Mozilla). I reproduce it here:
Essentially, the Omni implementation scales better with a large number of tabs. This is the first great improvement to tabbed browsing that I have seen in a long time. I can definitely see myself $30 for this thing.
They look quite weak from that video. They aren't nearly as useful as regular tabs (a la Mozilla, Safari et. al), as a matter of fact Omni's take on "tabs" reminds me of a glorified "Page Holder" from IE 5 for Mac OS 9. My real question is what does OmniWeb have to offer? It's using Apple's WebKit last time I checked, so it's not a new rendering engine that they have to give, no *REAL* tabbed browsing (which is what I want) so no dice there. I understand more choices are good yadda yadda yadda and all that jazz, but my question still stands, what does OmniWeb have to offer the end user?
Are you secure enough in your masculinity to run 'man touch'?
I must say I am impressed... I had never thought Opera would be eclipsed in cool features, but there it is. Many of these things Opera should be doing right now.
Workspaces, for example. Opera has an integrated system for easily saving and restoring web sessions, and even features an undo for closing windows (yay!). But this feature is buried in a menu somewhere, requires an open / save dialog box, and generally could be a lot more intuitive. Despite having been in several iterations of the browser, few people have found it.
Site-specific preferences. People have wanted this for a long time now, and I'm glad to see someone is implementing it. Pity it wasn't Opera. Opera supports preference sets, and many of them contain site-specific information, but in no way can all preferences be set on a site-specific basis. From the description it sounds like you could, for example, set your Slashdot home page to be your user page. I may be reading this wrong... only February will tell.
Adding searches... This is just plain cool. While opera allows you to use one of many pre-defined searches through a variety of means (including typing "g " + subject into the address bar), adding any search would be a powerful and useful ability. Of course, Opera's more flexible interface would have to find ways to deal with this (an individual search bar? the agregate search bar? the address search method?), but it shouldn't be too difficult.
Sharing bookmarks on a LAN is both great and troublesome. How do you implement this easily and quickly in a Windows environment without Rendezvous?
Tabs aren't as big of a deal, honestly. Usually either you have few enough pages open that you can keep track of them by name, or you have so many open that thumbnails would be too cumbersome to use.
I've always been envious of OmniWeb's History Search ability, website update notifications, and inline spell checker. That latter is being addressed in opera 7.5, along with a few nifty other features. While I will continue to use Opera, not the least of which because I have a PC, OmniWeb appears to be shaping up to quite the must-have app. OmniWeb was originally slated to ship as the default browers for OSX. Now it looks like that was a great idea.
The ______ Agenda
Apple does seem to have gotten sloppy with terminology once again. They can't call a component "JavaScriptCore" -- technically and legally, "JavaScript" can only describe the Netscape implementation of the language. The generic term is ECMAScript. Anyone taking bets on how long before Time-Warner's lawyers notice the trademark infringment?
There's a lesson here for those of us stuck with Gecko, Opera, or the mysterious combination of undocumented engines that is Internet Explorer. You want standardization, you gotta have open-source components. W3C puts a lot of work in defining standards for HTML, CSS, and SVG. These standards have a lot of unbelievably cool features, with much more in the pipe. But nobody can use most of them, because they're not widely implemented. What's the point of working so hard to create good standards if nobody uses them?
We need a reference web engine that will drive standards-based web development, just as the reference implementation of Java, with all its flaws, drove the adoption of the Java platform. Microsoft probably wouldn't use it, but it would provide some small pressure for them to be more standards compliant. W3C could develop such a comonent from scratch, or they could use Gecko; but KHTML seems to have the code base that's closest to a real tipping point.
Sigh. Thumbnails are optional. You can just show the window titles in a vertical list. Watch the movie.
i'd hit it so hard, if you pulled me out you'd be the king of britain [bash.org]
My three reasons for sticking with Safari are:
1) The Google Toolbar (although not implemented in the same full and correct way as the real thing on Windows).
2) Ad blocking
3) Pith Helmet - it allows ad content (or really any content) in a web page to be blocked. So banners and images can be stopped and not downloaded - saving my slow connection from having to bother with them, as well as not even seeing the ads.
I also like the bookmark bar, but I suppose many of the browsers have that now.
I know little to nothing about OmniWeb, will have to check it out more.
There are some odd things afoot now, in the Villa Straylight.
Actually one of the developers has been saying that he bas been working a lot on the program on his 12" iBook, and has been saying that both the tabs and the workspaces help him better manage his screen real-estate. Of course that means he was also using XCode on a 12" iBook, and you have to be a contortionist to do that well....
OmniWeb 4 was actually pretty stable and rendered fairly well even before they switched to using WebCore. The only thing that was really missing was some of the CSS support and some of the JavaScript stuff. The main problem the Omni Group had was that they were spending a lot of time chasing sloppy coding that worked in Internet Explorer because people only checked their code in Internet Explorer or they coded to some of the quirks in IE and that broke the rendering in other browers.
The Omni Group finally realized that they were trying to master too many disciplines and they were spending time re-inventing the wheel. They made a smart choice and decided to let someone else worry about rendering web content while they concentrated on solid UI and application design. The merging of the Omni Group's great UI-sense and KHTML's excellent rendering is a dream and they combine to make a wonderful product that is well-worth throwing a few bucks at. Not to mention that you can use the product indefinitely for free and all you'll see a single funny nag message out of a series of a dozen or so every few days.
Sapere aude!
Safari is not playing leap frog or riding on the tails of anyone else's efforts. Omni should first attempt to get their own browser out the door without crashing all over the place; when they've demonstrated they too can write solid code, then they can do what they want.
You're so right. How dare they build on open code, implementing open standards?!? These guys are hijacking the web, pure and simple!!!
Omni should first attempt to design their own markup language and transfer protocol. Once they've demonstrated they too can write a good network stack and lay their own backbone fiber, then they can do what they want.
Anything short (including their sickening use of public roads to drive to work!!!) is just sheer piracy on their part.
Timeo idiotikOS et dona ferentes
But doesn't Safari do just that? Apple didn't write Safari's rendering engine from scratch. Apple chose to use WebCore, which was derived from KDE's KHTML code. They've been very good citizens by contributing code back to the KDE project, but they're still utilizing code that was developed by another group.
So why is it okay for Safari to "leap frog" and build upon the efforts of the KHTML development team, but not OmniWeb? What makes Safari an "honest effort" when they re-use code in the same way as Apple? Why should OmniWeb have to re-invent the wheel when there is already a stable rendering engine available for them to utilize?
If you want a web browser with a "niche market" that uses a proprietary rendering engine, look no further than Opera or Internet Explorer. But the re-use of WebCore makes OmniWeb a credible product, and frees up their developers' time to do other worthwhile things, such as fixing bugs and working on the stability of their program. Why should they have to waste time writing their own rendering engine?
I can understand your dislike for OmniWeb for their past track record of stability, but if you're going to critique their re-use of open source code, you'll need to come up with some far more convincing arguments.
Super ninja monkeys will one day rule the world!
I believe that, coming out of the whole NeXT tradition that invented the web browser, OmniWeb used a real SGML engine to parse the HTML applications that are called web pages, as you could tell by reading the status bar as pages loaded (SGMLObject parsing, or something; I've since updated to OW 4.5, with WebCore). True, this engine had trouble with many pages--that is, it did not accept as liberally as it could have--but I think that the twists and contortions that people put HTML through would have made even Jon Postel hang his head in shame.
He was a verray parfit gentil knight.
At least in early versions, OmniWeb tried using a 'pure' object-oriented approach to render web pages. They wrote an object that converted HTML into SGML, then another one to convert SGML into RTF -- which was a native format for NextStep to display. Although elegant in architecture, these many conversions delayed the display of the web page quite a bit.
To make matters worse, until OS4.2, the low-level Next-supplied Text Object was not multi-threaded. (No real reason for it to be, until web browsers came along opening up multiple http connections and wanting their pages rendered ASAP.)
I believe if you get just the right combination of OpenStep 4.2 on a Turbo Cube, some rare EOF patch, and the right beta of OmniWeb, the whole rube goldberg contraption works just dandy. Otherwise, it's probably a good thing they are just outsourcing HTML rendering to WebCore nowadays.
There's some more info in an old post of mine here.
And we shouldn't forget. Omniweb was there first. Back when OS X was just out, there was only the hideous hideous IE5 (tragic since IE5/OS9 was the best browser of its time) and a nightmarish Netscape. OW was gorgeous and fast and showed everyone what a browser really should be like. Yeah, it had occasional crashes, but it was the best of the bunch. Except for the sick task of reverse-engineering all of IE's bugs. Leading to...
That's right, why use a standards-compliant free Apple standard Toolkit, when you can completely reinvent the wheel for no good reason other than to impress about ten people in the world who probably won't even register your product and will complain about how your URL field isn't adjusted right in your demo movie. While their at it, they should have written their own compilers, the cheating bastards.This isn't macho CS major dueling here. This is about creating useful software. Most of us are delighted they are using Webkit for rendering and thinking about other interface issues instead of wasting their time on rendering (which incidentally was killing their development efforts).
If you don't like it, don't use, or don't try it. But please don't try to say it's somehow dishonest or wimpy to use system toolkits.
MS had dropped development of IE for Mac OS X since June of 2003.
8 ,0 0.asp
http://www.pcworld.com/news/article/0,aid,11115
-B