Monday saw the somewhat quiet release of Safari 5. One can speculate as to whether widely-discussed network issues during the keynote led to Steve dropping the subject, whether some features, in his eyes, lack the polish to be worth presenting, or whether it had never been intended to be featured at all.

In any case, it is a worthy upgrade, bringing us Chrome-like DNS prefetching and geolocation, greatly improved address bar autocomplete (though no merging of the address and search fields — personally, I’m undecided, but many consider this a flaw) and the ability to undo the accidental closing of a tab. Consistently with just about any other application, this is accomplished through Edit → Undo, but other browsers would likely place this feature in the History menu, where even Safari itself has the related Reopen Last Closed Window.

Though I’m in favor of them, the autocomplete changes will take some getting used to. In the screenshot on the side, how do you select the URL to change part of it?

And then there’s Bing support. Safari on Windows and iPhone (but, oddly, not Mac) had supported Yahoo! as an alternative engine for a while; now, the Mac and Windows versions both support all three big search engines. This sadly does not mean OpenSearch support; unlike IE 7, Firefox 2 and others browsers, Safari still doesn’t support this Amazon-created standard for search engine information. It does, however, apparently mean that search engine configurations are no longer hardcoded in the binary itself; rather, there’s a ~/Library/Safari/Configurations.plist.signed enumerating the engines by name, home page URL, URL formats for queries and the respective suggestions API, etc. Past the end of the markup, there’s an RSA signature, suggesting that Safari won’t accept user modifications for whatever reason. Instead, cursory checks suggest it ignores the file’s contents entirely.

---

The two big features, however, are clearly Extensions and Reader.

Extensions

Credit to extensions as we know them today largely belongs to Firefox, which started off with a very simple, lean feature set (much like Safari 1.0, and in stark contrast with Opera at the time), but allowed virtually unlimited customization and enhancements. This was facilitated by Firefox’s cross-platform chrome¹ being developed in XUL, CSS and JS, much like a web page, except with an XML dialect for user interfaces rather than for the Web. This means relatively easy adjustments to just about any part of Firefox’s chrome.

Google Chrome, using native UI instead, followed the different approach of allowing little “web pages” to be written that hook into predefined APIs, such as to add another button to the toolbar. This yields less flexibility, but accounts for most common cases.

Finally, Greasemonkey, itself an extension for Firefox, allows JS-based customization of the content of web pages — e.g. removing ads, tweaking the layout or adding features. Chrome allows most Greasemonkey to run and treats them seamlessly as extensions.

Safari’s model is much closer to Chrome’s, though it has no built-in support for Greasemonkey.² Apple provides a Conversion Guide which explicitly mentions all three previous types of extensions. As much as I wish they had made Greasemonkey extensions just work, the feasibility of which is demonstrated by Chrome, but the conversion is rather easy, and the cool Extension Builder helps, too.

One oddity is the requirement to sign up with Apple’s Safari Developer Program. While free, it requires an account, and as of five days later, they still have yet to approve mine³. After that, you get to sign your extensions, and only then can you actually use the Extension Builder in a meaningful way. Perhaps Apple plans to use this facility for a kill switch, i.e. to weed out malware. Who knows — perhaps they even intend to run an opt-in Extensions Store.

Extensions, once ready to install, are .safariextz-suffixed xar archives. Unpacked, they’re .safariextension folders, with an Info.plist and various other files. The archive itself appears to contain a special tag (perhaps a digital signature assigned to the program member) that makes Safari accept it; simply recompressing the folder using xar -cf will not work.

Overall, the feature doesn’t seem finished. Enabling extensions is supposed to be possible through Preferences, but it takes a trip through the Develop menu (which is normally hidden). Only then (and, in my case, after another relaunch) does the entirely Extensions pane in Preferences appear, and even after that, I’ve had a previously-installed extension suddenly vanish entirely. Rather than directly offering an installation, Safari will download an extension, then require you to open the resulting, otherwise useless file.

Even the built-in update mechanism has its weaknesses: leaving the Install Updates Automatically checkbox unticked, I would still have expected to be notified of available updates. And once they were installed, they were also re-enabled if previously disabled, and toolbar items re-added if previously removed.

An Extensions Gallery is promised for later this summer. Until then, this Tumblr account does the job just fine.

Reader

If Extensions are finicky but useful, Safari Reader is even messier, and often useless.

Going by the revised Acknowledgments, the feature is largely based on Arc90’s Readability project, which I’ve been increasingly using in the recent months, as website layouts get less and less, well, readable. (My settings: Athelas, Large, with a Medium margin. Gorgeous.) When about to leave, but wanting to read something on the go, I instead use Instapaper, invoking its Read Later bookmarklet using cmd-7, then using Instapaper Pro on the iPhone to get a nicely formatted, distraction-free, no-fingers-necessary reading environment. Both Readability and Instapaper remove most of a website’s layout, focusing largely on text. For the desktop, I find Readability’s output slightly better.

Safari’s is… different. Some pages that work just fine in Readablity and Instapaper just don’t at all in Safari Reader, not offering the little Reader button in the first place. Others only offer a portion of the article, stopping at a seemingly random location of the page, without any indication to the user that there’s more to read.

Yet others, however, work great. Consider this four-page Computerworld article, which it excellently lays out all in one scrollable pane, with little separators signifying page breaks. It’s awesome for that, and neither of the two others offer this capability.

But alas, the results are really hit-and-miss. Of this two-page eWeek article, it only shows the first page, and then the comments — with a heading in dark blue on darker blue, no less. Clicking the Next link does get you to page two, but leaves Reader; you have to re-enter it.

Readability and Instapaper are both continuously improved. It’s unclear to me if Safari Reader updates itself over time, or if newer versions will only be supplied as part of general Safari updates, which would suck — because right now, it just isn’t that good.

Post scriptum

The unpolished feeling of the two big additions makes me wonder why Safari 5 was released at all. As a technology preview, sure. As a final product worthy of a marketing page, though, it almost feels as if Apple was trying to steal the thunder of the first final release of Chrome for Mac.

1. A.k.a. all portions of the user interface that aren’t the content itself. ↩
2. There’s always GreaseKit, a port of Greasemonkey itself. ↩
3. It also comes with weird “Once you’ve completed your purchase”-type screens and mentions of “iPhone”, even though it is free and unrelated to their recently-rebranded iOS platform. It’s clearly a quick and dirty derivative of their other two programs. ↩

• WebM is now BSD-licensed. Good. (via @timbray)

• Hawt. (via @rands)

• Belgium’s first baby snow leopard. Sure, but does it run Safari? (Ba-dum-dish. via Denis.)

• The new trend: tweaking your PC from your iPhone. Asus shows off BIOS settings; Galaxy overclocking. (German article)

• Puntastic. (via @uliwitness)

• If you e-mail Steve Jobs, he may give you terse, blunt, yet useful answers, even the second time. If you e-mail AT&T CEO Randall Stephenson twice, you get threatened with a cease and desist. What about Valve’s Gabe Newell? You’re good.

• Apple has made a HTML5 marketing page, including a showcase of various technology demos, and I never knew they had sample code like that. Good, right?

  

  Yet for some inexplicable and likely stupid reason, they decided to not just recommend you look at the demos in Safari; they require it. That the “You’ll need to download Safari to view this demo.” error message can be worked around is besides the point. Not only does this still break cases where CSS vendor prefixes are relied upon; Apple makes little effort to support Mozilla’s and Opera’s respective prefixes. It also appears to disagree with the message of their marketing copy, which promotes “web standards” which are “open”.

  Requiring one particular browser doesn’t sound very standard or open to me. These are all very cool demos, and counter the idea that Apple wants to lock everyone into Cocoa Touch — but the browser lock-out leaves a bad aftertaste. (also via @timbray)

  Having said all that, there’s a sarcastic image going around showing Google Chrome doing significantly better at a HTML5 test site than Safari (142 vs. 70 points). However, I’m getting 120 points with the current Safari release 4.0.5, and 144 with the current nightly build. Since I find it unlikely that Safari on Windows fares that much worse, the screenshot must be of a significantly out-of-date version (yet, curiously, a not-so-out-of-date Chrome). Counting Apple’s unfair play with yet more distorted comparisons doesn’t benefit anyone.

• Google Android open source and compatibility program manager Dan Morrill:

  Because [fragmentation] means everything, it actually means nothing, so the term is useless. Stories on “fragmentation” are dramatic and they drive traffic to pundits’ blogs, but they have little to do with reality. “Fragmentation” is a bogeyman, a red herring, a story you tell to frighten junior developers. Yawn.

  I can think of multiple meanings of “fragmentation” as it applies to Android, including differences in:

  • hardware configuration, e.g. GPU capabilities,
  • hardware properties, e.g. native screen resolution and hardware buttons, and
  • minimum and maximum supported operating system version

  — so the point that it means a lot (though far from “everything”) is well-taken, but this only makes matters worse: every possible meaning I can come up with gets the very point of pundits across: it’s bad, and to a point, it’s unnecessarily bad.

  I’d be the first to support advances in technology and ditching legacy components¹, and that will inevitably lead to some users getting left behind. But as an Android user, you can’t even be sure that a handset you buy right now will offer the best possible hardware prerequisites, and run the latest version of the OS.

  Call it a frightening story all you like. It’s also very real. We’ve seen this before with Windows Mobile, and if Google doesn’t step up to require hardware manufacturers and carriers to make the experience less awful, we’ll see it again. (via proggit)

  More on this by a Slashdot user. (also via proggit)

• It may be a bug, but double-/triple-pressing the lock button apparently invokes a flashlight in iPhone OS 4.0 beta 4. (via fscklog)

• This appears to be the year of Star Trek stuff getting real. First, the PADD gets released by Apple; now, Pantec Biosolutions gives us hyposprays called P.L.E.A.S.E. (via @alahmnat)

• This guy must be the Poe’s Law stand-in for libertarians: “Why Not Feel Sorry for BP?” (also via @alahmnat) Ooh! Ooooh! I know a reason. (via DFLL)

• Wikipedia has a List of youngest birth mothers (and grandmothers). I for one was not expecting those ages.

• A Services Menu for iPhone. Great (and, evidently, Apple-esque) approach, but doesn’t that require a way to compile bundles (that aren’t apps) on the iPhone, and ship them through the App Store, perhaps alongside apps? After all, you wouldn’t want the whole application to launch… (also via DFLL)

• Some suggestions for Xcode UI Improvements. Very cool UI design, and looks like a bunch of huge time-savers, but would probably take a ton of effort, or be very inconsistent / cursory (especially, say, Brandon’s in-API-doc example screenshots of enum values). (via @sophiestication)

1. Why can I still not order a MacBook Pro without an optical drive? ↩

• Adobe Announces iPad-Focused “Digital Publishing Platform”. Generating HTML from Word? Pshaw. Let’s generate entire iPad apps from InDesign.

• David Baron on Hidden complexity in specifications.

• Kaleidoscope, a new file comparison utility for the Mac. Looks gorgeous, and image comparison is a useful addition — but with “it doesn’t support editing or even applying changes” and “it doesn’t do folder comparisons”, Michael Tsai points out two jarring weaknesses.

• Because, really, why would you expect Photoshop CS5 to handle something that 1.0 could?

• Onkyo showed off a notebook with two displays; one optionally sliding out from beneath the other. Ten inches and 1366×768 pixels each. The displays aren’t flush next to each other, and the (German) article doesn’t speak highly of the build quality. Puzzling that so few companies (such as Apple) ‘get’ seamless hardware design.

• A quote about engineering. I’d argue that, compared to building bridges, software development is much fuller of unknowns, but this may merely present my ignorance in architecture.

• Just one question. Also, how many cubic inches is an ounce?

• Is this Guid HOT or NOT? (via @codinghorror)

• “Will Apple Embrace the Web? No.” No indeed, because they already have, long ago, being a WHATWG founding member and openly leading the development of what is widely regarded as the most impressive layout engine du jour, WebKit.

  Camen likening Safari to IE6 might make some sense if Apple had a controlling market position for cellphones (2.7%), or at least the smartphones segment (15.4%). They’re not even in the first or second place, and chances are Android will soon surpass them. Apple may control the mindshare (and has been leading in profits), but they’re far away from leading the market, and don’t seem intent to change that.

  Do you remember when Internet Explorer 6 was released? It was hands-down the best, most standards compliant browser at the time.

  In some bizarro universe, that may be true, but Internet Explorer 6 was released more than a year after Internet Explorer 5 for Mac (yes, by the same company), based on the Tasman engine, which offered much better support for CSS, PNG and various other aspects.

  Can you see where I’m going with this?

  No.

  Well, actually, thanks to your mention of John Gruber and claim chowder, my money is on ‘linkbait’. I fell for it. Good job. (via proggit, where this kind of drivel does not belong)

• I really want to do something nifty with CSS3 media queries, like the example(s) laid out in Responsive Web Design (via Jon Hicks), or the for-IE9 demo by Microsoft.