waffle

What I don’t understand about the normally healthy “new version of Windows” skepticism is how it’s being applied to Windows 7.

Windows Vista was not as good as anyone would have liked it to be. It came with what is probably the toughest relative demands on system requirements of any version of Windows at the date of release. It took forever, including one more or less complete restart, to develop. It used too many system resources, something that the service pack was unable to conclusively resolve. It introduced some well-needed changes, but in a way that broke far too much backward-compatibility. Most importantly, more than any previous version in history, for whatever reasons, it made people downgrade to what they had before.

All of these things are resolved, at least to some extent, in Windows 7. You’d think that would be enough that people would regard Windows 7 as better than Vista. If not that, then why not the lower demand on RAM (if you get Aero Glass, that’s 128 MB more memory freed up right out of the gates) and CPU in actual use? The only significant cons that I can find in Windows 7 has to do with the bone-headed decisions to keep most of the editions, price them too high and to disallow upgrades directly from Windows XP.

I get that some journalists sold Vista as more effective and profound than it really was and are now cowering a bit. But the numbers are in: people have, in real life, responded far more positively to the pre-release versions of Windows 7 than to the corresponding Vista versions (and for that matter to Vista RTM and SP1), going so far as to claim they should just ship the beta. I don’t think anyone would have made that demand for the early Vista pre-release.

Windows 7 is not a good release because Microsoft needs it to be, or because Vista lowered the standards, or because of marketing. Windows 7 is a good release because it’s a stable piece of software in its own right that’s the best Windows yet in every regard, and that’s better than Windows Vista in all the right ways and better than Windows XP in all the ways that Vista should have been, but wasn’t. I just wish we could give them credit where credit is due, all sniping aside.

I’ve been unlucky enough to have held the same opinion about Twitter ever since it was launched. This opinion is firmly entrenched in one basic, fundamental problem.

The problem does definitely not have anything to do with scaling; I said that my opinion had been the same, and scaling’s varied.

The problem with Twitter is not that people may write whatever they want. The problem is not that people may not care. The problem is not that it shouldn’t be more than 140 characters, nor that this breeds (short) one liners, nor that it blocks the great American novel from being published in an entirely inconvenient form factor. The problem is not that URLs are folded up into basically URL pointers. The problem is not that previous entries are mindlessly copied verbatim under the guise of “quoted for truth” without any headroom for clarifying comments (they can be added later, and this is a problem in any form of writing).

The problem is not that the briefness of the form ostensibly breeds a polarized, nuance-and-information-free variety of criticism (this is more of an issue about time constraints and a willingness to familiarize ourselves with certain points of view). The problem is not that we could do better things with our time, like complaining about how other people are writing instead of spending the time to complain about how other people are writing.

The problem is not that it or its users can’t make up its mind about whether it’s publishing or communication. The problem is not even how everyone and their fucking cat, because of that, has a new theory for how Twitter is the newest step in human evolution and communication and how this benefits you and your company.

Some of those things above are actual issues that make me shudder and may keep me away in the future. But they are not the problem.

The problem with Twitter is that it’s got a thoroughly shitty user experience.

It works reasonably fine for reading just regular tweets, reversely chronological as it may be. But because almost everything is either a node in a conversation, a node in a subject, or ripe to turn into either or both, actually following that thread turns into a wild goose chase. You have to read everything top-posted and manually unwind the conversation while glimpsing every other or third or fourth reply while doing so. And heaven forbid you want to follow a back-and-forth that’s on the third page; you’ll have to go back and push “more” twice before you can even retrace your steps.

Twitter might be one of the most inconvenient methods of following a conversation in recorded history, and I can’t understand why everyone else seem so willing to go along with it.

Some people may want to point me to Quotably. Quotably failed because if you’re trying to do something that requires you to cross-correlate everything, you better have an efficient handle on everything, and they didn’t. Some people may want to convert me to one of countless Twitter clients. I haven’t seen this problem handled ably in a Twitter client yet; I’ve had no impetus to actually download and use one because there’s been no evidence that the problem has been tackled. (If it has, you’ve got to work on your screenshots, people. Crack this one and the world is your oyster.)

Some people may want me to get an account and start keeping track of people. No. I don’t want to “follow” everyone to opt myself into a better UI, and I have no interest in publishing. What kind of bullshit is that?

So I’ll do what I’ve always done. Occasionally look up those whose opinion or writing I care about, rummaging through the rabbit droppings in a desperate search for chocolates (vivid imagery appears courtesy of Zero Punctuation), by hand. Silently fuming.

• Is there something tremendously wrong with me, or are none of these actually funny?

• Peter Norvig writes a “did you mean…” spelling corrector in 21 lines of Python (2.5). At the bottom is a list of ports along with lines of code: the only Java port to date weighs in at 372 lines, more than double that of the C port (184) — which, I must remind you, has less string smarts than Java — and 3.8 times bigger than the C++ version (98).

  I don’t hate design patterns or carefully architected classes and hierarchies, quite the opposite, but there’s such a thing as getting in the way of solving the problem. Someone needs to write “The Cognitive Style of J2SE”. Oh wait, sorry.

  And I’m going to shove the C# port line count (22) in the face of the next person who tells me that “C# is just a poor Java clone”. Then I’m going to stop listening to them. These people think Java of 2001 is so perfect that not only will C# forever stay an inferior clone of that snapshot, Java itself should, as well. They tried progress once and they didn’t like it. The success of the Java platform now hinges on how willing it is to finally ditch the flagship anchor in favor of something better.

• More stylistically poor reporting at AppleInsider. Here’s what I take away from the headline and the preamble: Apple inherited ZFS from Sun and are now the sole master of its destiny (CDDL forkability aside) and they’re discontinuing it. Here’s what actually happened: Apple’s port of ZFS to Mac OS X was shut down. The issue is eventually reported in clear enough language that you’d know… below the fold.

The only decent Help experience I’ve had with Microsoft developer tools involved my first experience with Microsoft developer tools. Maybe it’s the misty glasses of nostalgia, but from what I can so vividly remember, in Visual Basic 3, 4, 5 and 6 you could select anything, hit F1 and get a good rundown of whatever that thing was. This wasn’t spectacular, but it reminded me of good shoes: you don’t notice them.

Since then the general trend has been steadily downhill. The initial .NET versions of Visual Studio had a horrible Help experience where the thinking was that with all those palettes (I’m sorry, “tool windows”), what could go wrong if we added some more? Also, if we made it slow? Horribly so? The 2005 version cleaned up a bit by moving everything into a separate window; however, in a classical case of everything that Microsoft is doing that is wrong, unholy and “enterprise” showboating, this was in fact a “Document Explorer”. Also, still dog slow.

2008 — where we are today — is not as bad. The speed is up, which counts for a lot. But the language selector won’t remember your settings so apparently, I want that signature in VB (declaration), VB (usage), JScript.NET, J#, C++/CLI, F# and Brainfuck, and I want all of those variations to appear ahead of those in C#, and I want the syntactic differences to merely be hidden in the DOM rather than eliminated so that when I copy something, I get “string..::..Join” because hidden text is still included. And let’s not talk about help merging and one incorrectly run instance can force a full reinstall, or Microsoft’s 1999-era notion of “search is either fast and indexed on keywords or slow and a tax form”.

So, yeah. I wanted more for 2010, which is why I was hopeful when they ran a survey a year or so back asking for general feedback on the Help content and on the Help experience. Add to that the sudden outbreak of common sense when it was decided that they’re just going to zip a bunch of HTML files and an index and call it a day, and the promise of a new viewer written in WPF. (WPF itself isn’t important and sure as hell isn’t going to improve on a bad design, but it forces a rewrite (and thus a place for reconsidering the design) and is conducive to virtualization in, say, an index list or a big fucking table of contents tree.)

What we get instead is a small, fast, purpose built web server that runs in the tray and serves up everything in the default browser. Not only that, this is an actual, undoctored screenshot of some sample code:

Enough already.

• Opening everything in your default browser is a huge inconvenience, not to mention that the page is broken in Chrome.
• Searching brings up a separate page for hits instead of listing APIs in-line (suggest-style) and perhaps showing prose articles with summaries. Not only is it 2009, your company actually invented AJAX and claims to run a major search operation.
• The idea of a new viewer was thrown around until very recently. What happened to it? A viewer is the right way to go and the only way to implement bookmarks, table of contents and keyword indexes without needing to dirty the browser or half of the page (which will take time, space and resources to load).
• A small palette in Visual Studio to show more of the discussion around the method (exceptions thrown, more detailed information about what the method does; what you don’t see in Intellisense, basically) would be a great addition. As a popup, it sure was great in Netbeans several years back. Are you going to be shown up in user experience by a Java developer tool? (That’s not a “.NET vs Java” thing, that’s more of an “embarrassment” thing.)
• The cleaner MSDN Library variety that debuted a few months back was great except for the small-ish font and insistence on no scripts. That means that every signature shows up in every supported language; no hiding whatsoever. The Help version chooses a better, bigger font and stuffs the signatures in a “tab view”… which defaults to the wrong language (VB) and whose default tab you can’t change. Sigh.

There’s no shortage of good intentions. I’m willing to accept that VS2010′s Help may on the whole turn out a net win when compared to VS2008′s (for example, the new interface for downloading more documentation is much appreciated and the removal of the tedious and dangerous index merging shouldn’t be underestimated). But that doesn’t mean that there aren’t glaring flaws; flaws that have been obvious for years, and freshly minted ones that are just as glaring.

The fact that it’s taken so long to get it this right, even when this right is spectacularly wrong and incrementally worse in so many aspects, speaks lowly of Microsoft’s level of ambition and priorities. I know that that draws some snickers from the regular audience, but I’ve seen enough on other fronts to know that people inside Microsoft actually care about what they contribute. I would like nothing more than for Microsoft to please put some tools and resources in the hands of those who care about Help, instead of, like so many other times, not and saying they did.