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.

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

And so, I’m writing this post on a mobile platform of Apple’s both famous for its revolutionary UI improvements as well as infamous for its (often deliberate) closedness and restrictiveness. Can’t a man have his porn?

I only have a 3G, which isn’t among the fastest of the bunch any more, and by the looks of things will feel even more obsolete in a few weeks’ time come WWDC, despite being a mere 13 months old. Still, I regard it as a wonderful companion despite the technical and social (i.e., put forth largely by Apple and the carrier) barriers.

I’m writing it in Notes, complete with Marker Felt, screen-wasting UI, zero remotely blog- or markup- or even editing-related features beyond a keyboard and a text view and other properties that make it largely unsuitable. I do so because my initial (and far more obvious) choice of app, WordPress, proved even worse, and finding something better will probably prove to be a time- and money-consuming process, perchance not even netting me a winner.

Though I had never really used it, WordPress was among the first apps I installed. It is unusual in that it proves the notion wrong that you cannot have open source apps in the App Store; I have never heard of Apple so much as threatening to take it down or asking for changes. Its basic interface is also well-structured, and setting my blog up in it was a snap. And it offers nice newbie features I may or may not ever use, such as picture upload.

But the very nature of iPhone apps dictates self-containedness; that is, switching to a dedicated editing application, as I have ultimately done, is too awkward to be the default assumption. So built-in editing better be good. And it’s not. It either far more limited than one might reasonably expect, or so unintuitive — nay, undiscoverable — that the “features that can’t be found might as well not exist” rule applies.

• I can find an HTML preview, but no markup editing controls. three buttons would save tons of time: Wrap Selection in Tags, Wrap Selection in Tags Per Paragraph (especially for <p> and <li>) and Close Current Element. None of this appear to exist; in fact, I couldn’t find any indication that anyone had ever tried to write a post with more than absolute bare-bones plain-text formatting using the WordPress app.
• Working with the post up to this point crashes it. Every time. This isn’t a 1.0. It’s a 2.4 as of this writing. (Somewhat oddly, it won’t update 1.x versions; no explanation is given anywhere that I can see, and due to the mandatory sandbox model, this means you lose all set up blogs and local drafts, as far as I can tell.) My post is hardly long or complex, so I can’t think of a reason why. Either way, that literally makes it impossible to work with.
• It informs you of “recovered” changes, but still inexplicably loses data. One of the user-friendly changes iPhone OS has brought us is largely eschewing notions of saving files or other forms of records. Whatever you do is just there; you don’t have to care (but also cannot influence) where. The WordPress app defaults to local drafts (it’s not entirely clear to me why it has those at all; with the exception of an iPod touch and non-3G iPad, you can and should assume connectivity, and transparently save to what hipsters call “the cloud”), which means you can get going almost instantly. Good. But when returning to it without having previously expressly tapped the “Done” or “Save” button, it’ll notify you that it has restored your unsaved changes. This needs no mention; it is expected of virtually any iPhone app. Despite this dialog, the changes apprently don’t immediately get merged into the draft you were working on, presumably as an undo fail-save of sorts. It just caused me to lose all my changed due to user confusion. Not cool.

I expected some sort of help, such as a built-in Markdown preprocessor. I can think of a lot of funner things to do than writing raw HTML tags on a virtual keyboard on a 3.5-inch screen, and I can only assume other users feel the same.

So, my current editing interface is Notes, mostly by virtue of being fast and getting the job done. As for posting, I’ll try WordPress’s regular admin web UI; it doesn’t come with a mobile version, and previewing posts appears to be broken on my site (even when I try with the default theme), but it absurdly enough feels faster than the dedicated app. Typing into form fields in Mobile Safari, on the other hand, continues to be a nightmare and not even worth considering.

So much for editing the post. What about preparing it? That situation is actually even worse.

Chuckellania, by its very nature, is a collection of neat stuff I found all over the Web while using it (or other parts of the Internet). Sources include, but are far from limited to, news feeds (RSS, Atom), tweets and other private or public messages (Twitter, AIM, MSN, GTalk, Skype, Facebook, IRC, e-mail and others), forum posts, and of course my own browsing history. This presents a logistical problem: how do I not only mark and/or rediscover links of interest, but also channel them together, with attribution preserved?

I like how constraints and other factors have made the iPhone OS platform rife with site-specific apps. It has its downsides, but it also allows (and oftentimes has made) developers to create an impeccable experience. No platform — not the Mac, not Android and certainly not Windows — has sparked such an amount of continuous design innovation and excellence in Twitter clients. And it’s a subject for a separate post, but Instapaper Pro on the iPhone is perhaps the best reading experience for articles on the Web that I’ve seen anywhere. It’s streamlined, focused, non-distracting, and you get to scroll by tilting the device. People that don’t get WriteRoom’s or OmmWriter’s raison d’être won’t get this either, but I think I can live with that.

But walled garden arguments aside, the thing with having separate apps for sites is that inter-app exchange of data becomes necessary. Some developers recognize this. It’s why you often get “Send to Instapaper” buttons here and “Open in Safari” menu items there. Other times, such integration is either entirely lacking, or too limited for one’s needs. I’ve found that, in many cases, the “Save for Later” workflow leaves much to be desired.

My Twitter client of choice is Twitterrific. It offers a built-in browser so that, when clicking on a URL in a tweet, there’s no need to switch apps. From within the browser, it lets you open the page in Safari instead, copy its link, e-mail or bookmark it, or send it off to Instapaper. For quickly exporting info from multiple tweets to another app, copying is useless — you’d overwrite the previously copied URL each time. The same, due to immediate app-switching, applies to the Safari-labeled button. The latter three all share the virtue of sending the info off while staying in Twitterrific.

However, that’s for a URL within the tweet. Omitting attribution of discovery is uncool and to be avoided, so what about tweets themselves? Tapping and holding onto a tweet lets you copy its URL — again, only useful for a single one. Of the options presented by the asterisk-labeled button, only favorite, mark and e-mail are interesting. While a nice built-in means of swiftly collecting tweets that have piqued our interest, with a beautifully simple filtering UI for finding them later on, marked tweets don’t appear to be available in any form externally, making the feature insufficient for my needs. This does not apply to the other two, but the lack of similar bookmarking and Send to Instapaper options at this location is unfortunate.

The best options appear to be to mark them, then either do the copy-Home button-Safari-new window-paste-reopen Twitterrific dance for each of them — a frustrating exercise on this slow device of mine — or to e-mail them to myself; Mail, at least, can stay running without resorting to hackery. Favorites would avoid the switching altogether, but be semantically incorrect to me; I regard a favorite tweet as something all-time and thus more significant than a mere item on Chuckellania.

NetNewsWire is next  Largely a synced offline front-end to Google Reader these days — but a bit comfier overall than Google’s own quite impressive iPhone-optimized Web interface — , it also sports in its own browser. Unlike Twitterrific, it offers a Send to Instapaper option for feed items themselves, not just browsed webpages. Thank you.

Admittedly, it doesn’t normally make a lot of sense to read a tweet in Instapaper. But the point is less to find an ideal combined “Inbox”, to use GTD terminology, as it is to find any shared one at all. If both T’ific and NNW offered “Send to Notes”, “Append to Clipboard”, “Add to Bookmarks” or similar choices, that’d be even better.

My IM client is Beejive. It’s great (otherwise, why would I use it?) — it uses a bouncer-style centralized connection with push notifications, supports quite a selection of services, has a simple, cute interface and hasn’t failed me much. It’s slow to launch an sometimes says messages haven’t been sent when they actually did arrive. But still.

Beejive also has its own browser. It lets you copy individual messages (and, thus, the URLs within as well as their context) through tap-and-hold, if you fancy that, but this is hardly good enough for attribution: who wrote that? When? And are they going to stab you in an alley tonight?

The process of aggregating item for Chuckellania is therefore much more time-consuming on an iPhone than it would be in a multitasked, multiwindowed environment such as Mac OS X. Some of this will get better in 4.0. Some is simply a matter of improvements in individual apps, or me choosing alternatives. Much, however, is a direct result of constraints. This platform wasn’t built for such uses. Will it ever be?   

Inspired by OkCupid’s message style (as seen to the right), I set out to try and recreate most of the effect without the need for any images.

Our goals:

1. A slightly rounded border around the quote’s container. (The above doesn’t actually have this, but this makes it slightly more speech bubble-like.)
2. A right triangle, with one of its legs right next to the quote’s container.
3. Alternating background colors.

In addition:

4. The speaker’s picture could be displayed, but we won’t explore that here.
5. The entire container including the triangle should have a subtle drop shadow, but due to the technique used for 2, I haven’t found a feasible means of accomplishing this.

Our markup:

<!DOCTYPE html>
<html lang="en">
<head>
	<style type="text/css" media="screen">
		/* to be filled */
	</style>
</head>
<body>
	<div class="post">
		<div class="triangle"></div>
		<blockquote>
			<p>Lorem ipsum dolor sit amet, consectetur adipisicing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est laborum.</p>

			<p>Vivamus vitae magna a sem dapibus faucibus. Phasellus sit amet erat ut quam sodales congue a quis erat. Quisque ultricies mi sed urna ullamcorper fermentum nec ut purus. Cras leo odio, pellentesque eget egestas vitae, mollis at purus. Nunc id arcu mi. Nulla facilisi. In arcu odio, convallis id vehicula vel, tincidunt vitae nulla. Nunc vulputate fringilla justo, ut bibendum magna gravida at. Duis eu mauris eu neque luctus cursus. Sed varius mi non velit pellentesque vulputate.</p>
		</blockquote>
	</div>

	<div class="post">
		<div class="triangle"></div>
		<blockquote>
			<p>Cras eget augue sed risus aliquam congue. Duis felis nibh, ullamcorper vitae hendrerit non, tristique eu leo. Phasellus vitae nulla ut magna malesuada aliquet sollicitudin quis massa. Curabitur tortor turpis, sollicitudin vel tempus ut, hendrerit a massa. Curabitur id velit venenatis erat blandit tempor eu et odio. Suspendisse in elit ligula, at gravida massa. Quisque pulvinar iaculis tempor. Cras ultricies turpis nec diam varius consectetur. Donec in lacus sit amet justo euismod tincidunt. Praesent dictum blandit varius. Nulla facilisi.</p>

			<p>Fusce sollicitudin laoreet enim, vitae mattis mauris laoreet et. Cras mollis rutrum neque ac varius. Suspendisse sit amet tortor ac odio luctus convallis vel aliquam dui. Maecenas vitae nunc ut sem condimentum eleifend et ut neque. Quisque vel metus mauris, nec fringilla dolor. Donec id sapien massa, vel euismod enim. Morbi ante eros, aliquam vel tincidunt quis, cursus id mauris. Integer id orci metus, sed porttitor nulla. Aliquam erat volutpat.</p>
		</blockquote>
	</div>

	<div class="post">
		<div class="triangle"></div>
		<blockquote>
			<p>Praesent auctor tellus at orci sollicitudin in tincidunt nisl suscipit. Ut pellentesque lacus id mauris lacinia tincidunt. Morbi metus leo, rhoncus a porttitor quis, tincidunt vitae neque. Aenean tortor sapien, ultricies non blandit quis, ornare ac leo. Nam volutpat justo sapien. Pellentesque in purus mi. Etiam ultricies quam placerat risus scelerisque cursus. Curabitur sed nisi massa. Etiam in turpis ac mi sollicitudin laoreet interdum a tortor.</p>
		</blockquote>
	</div>
</body>
</html>

This is pretty close to what markup containing several quotes should look like, with the exception of the triangle-classed <div>s. We’ll use those in a bit.

Thanks to CSS3’s border-radius property, the first goal is easy:

.post blockquote {
	-webkit-border-radius: 5px;
	-moz-border-radius: 5px;
	border-radius: 5px;
}

Neither HTML nor CSS have a native notion of shapes; arguably, they shouldn’t — that’s in the realm of SVG and other vector graphics formats. Still, requiring an external pre-rendered image or introducing another language just for this one shape seems excessive, so we’ll use some trickery. As explained in “Pure CSS Shapes: Triangles, Delicious Logo, and Hearts”, we can make use of the way ordinary borders are created — namely:

The corners of borders meet at a 45° angle, which is apparent with larger border width.

We’ll have to go one step further, though, rotating the triangle such that it fulfills our criteria. Luckily, we have CSS transforms:

.post .triangle {
	margin: 0 -1em 0 1.5em;
	border: 1em solid transparent;
	border-left-color: #bbf;
	float: left;
	-webkit-transform: rotate(-45deg);
	-moz-transform: rotate(-45deg);
	transform: rotate(-45deg);
}

Finally, alternating the background colors can be accomplished thanks to the nth-of-type pseudo-class, like so:

.post:nth-of-type(2n) .triangle {
	border-left-color: #77f;
}

.post:nth-of-type(2n) blockquote {
	background: #77f;
}

The entire CSS:

.post .triangle {
	margin: 0 -1em 0 1.5em;
	border: 1em solid transparent;
	border-left-color: #bbf;
	float: left;
	-webkit-transform: rotate(-45deg);
	-moz-transform: rotate(-45deg);
	transform: rotate(-45deg);
}

.post:nth-of-type(2n) .triangle {
	border-left-color: #77f;
}

.post blockquote {
	padding: 0.5em;
	background: #bbf;
	-webkit-border-radius: 5px;
	-moz-border-radius: 5px;
	border-radius: 5px;
}

.post:nth-of-type(2n) blockquote {
	background: #77f;
}

blockquote p {
	margin-top: 0;
}

And here’s what it looks like in a WebKit-based browser (in this case, Google Chrome).

Gecko isn’t so lucky: even in 3.7a4, there’s a slight rounding error (at least I assume it to be an error), causing a white line right through the center of the triangle, as if to draw the altitude.

Adding a shadow to the <blockquote> nodes isn’t a problem, but while I could get the shadow focused on the bottom left section of the triangle’s <div>, I couldn’t make it disappear elsewhere.

Still, it’s been a fun experiment.

Consider a UserControl PermissionRow that consists of three drop downs, and two buttons:

|Domain ↓| |Group or User ↓| |Access Level ↓| [Duplicate] [Delete]

Domain can be one of “Owner”, “Specific” or “Everyone”; only in the middle case is the second drop down visible. I.e., you want a UI to specify fine-grained access levels per domain, per group, or even per individual user. This array of settings is serialized back and forth in the database as a row, and displayed inside a FlowLayoutPanel.¹ An Add button underneath adds a new row. The Duplicate button creates an identical copy of the row for later editing; the Delete button removes it.

When loading from the database, a simple ORDER BY statement neatly places each domain together, and furthermore sorts groups within the “Specific” domain by name. As the user freely adds rows of their own, however, they will confusingly appear the bottom. Unfortunately, a ControlsCollection (such as our FlowLayoutPanel’s Controls property) does not appear to implement any form of sorting. It does, however, allow for manual changing of the index, i.e. the order, and will automatically take care of bumping adjacent elements out of the way.

Using an event, an implementation of IComparer, and a Generic.List, we can implement a simple sorting mechanism using an event:

Public Class PermissionRowComparer
    Implements IComparer(Of PermissionRow)

    Public Function Compare
    (ByVal x As PermissionRow, ByVal y As PermissionRow) _
    As Integer _
    Implements IComparer(Of PermissionRow).Compare
        If x.Sort1 = y.Sort1 Then
            Return x.Sort2 - y.Sort2
        Else
            Return x.Sort1 - y.Sort1
        End If
    End Function
End Class

Public Class PermissionRow
    Public Event SortChanged()

    Public ReadOnly Property Sort1() As Integer
        Get
            Return DomainChooser.Value
        End Get
    End Property
    Public ReadOnly Property Sort2() As Integer
        Get
            ' Depending on the domain, UserChooser.Value may not be meaningful; only use when visible
            If UserChooser.Visible Then
                Return UserChooser.Value
            Else
                Return 0
            End If
        End Get
    End Property

[..]

In the class hosting the FlowLayoutPanel, we need to add a handler for PermissionRow.SortChanged:

    Private Sub pRow_SortChanged()
        Dim pRows As New List(Of PermissionRow)()

        ' Add all eligible controls to a list
        For Each ctrl As Control In PermissionsLayout.Controls
            If TypeOf ctrl Is PermissionRow Then
                pRows.Add(ctrl)
            End If
        Next

        ' Perform the sort
        pRows.Sort(New PermissionRowComparer())

        ' Reorder the controls accordingly
        For Each ctrl As Control In pRows
            PermissionsLayout.Controls.SetChildIndex(ctrl, pRows.IndexOf(ctrl))
        Next
    End Sub

The big catch? No animation. I’m spoiled by Cocoa, in which smoothly animating the items of an NSCollectionView is not a problem. SetChildIndex, however, knows no such thing; the order will simply flip from one to the other. In my brief testing, this may make the user confusion even worse, as the row they were just editing is suddenly at a different position, without an obvious indication as to why. dot-net-transitions looks like an interesting (and also Cocoa-inspired) way to implement transitions for property changes, but this case isn’t one. I don’t know how the FlowLayoutPanel reacts to SetChildIndex works internally; there does not appear to be a ChildIndexChanged event.

In addition to rewriting the code in C# (or IronRuby, if you will), writing a suitable subclass of FlowLayoutPanel that could feasibly implement animation is left as an exercise for the reader.

1. Or, say, a TableLayoutPanel.

If there’s something lacking in Transmit 4, though, it’s release notes. (via Daniel Jalkut)

1. As an example, opening a 94 KB file through in an already-running TextMate through Transmit Disk takes 8.5 seconds; switching back to TextMate takes another 3.5. Working with any document in TextMate appears to significantly slow down whenever at least one ‘high-latency file’ is open, too. This makes edit in TextMate, test in other app, edit again workflows tedious and frustrating.
2. Keep your Duke Nukem Forever jokes at the doorstep.
3. Thanks to Jesper for the correction.

Discussions of or links to or the presenting of reverse engineering or exploits of any Cyan game (MO:UL and its variants expressly excepted) are not allowed. This includes tools used for the modifying of or distributing of copyrighted materials such as game executables, game data and game code.

[..]

Linking to webpages that break these forum policies for the purposes of getting around the forum policies is not allowed.

You might mistakenly think that, as the focus is on “hacking” MO:UL, rather than other Cyan products, this is okay:

As the plugins and code are released for MO:UL, discussions about the plugins and code are allowed for the purposes of learning and improving MO:UL, its variants (e.g., MORE, MO:ULagain, MO:UL Open Source), its plugins and its program code. The fine line here is if the discussion compromises Cyan’s other products (e.g., Uru: Complete Chronicles, Hex Isle, MQO), which is not allowed (see below).

Marten provides a good summary on why the above paragraph is essentially meaningless. The feasibility to actually use Cyan’s own released plug-ins is so limited it may as well not exist in a community sense:

As I understand it, the only legitimate tool available for Age writing at this time is the 3dsMax plugin. People who’ve tried to use it have reported that it is buggy, it is unusable without 3dsMax 7, and 3dsMax 7 isn’t sold anymore, and of course, we’ve been asked not to use the plugin without a legitimate copy of 3dsMax. Given these problems, it should be understandable if the GoW isn’t directing people to use this plugin, and instead continues to promote an approach that is not endorsed by Cyan.

Furthermore, there is no legitimate platform for testing and experiencing fan-made content at this time. But nobody wants to write Ages and not be able to link to them.

In other words, even if you and the other members of your pet project all happen to have 3dsMax 7, which cannot be legally obtained any more, you have bugs and limitations to fight with, and have absolutely no means of testing, much less deployment. The free, working, community-made and -maintained and easily available alternative, on the other hand, is not allowed. Neither is discussion thereof. Nor is linking to third-party sites containing discussion thereof!

For some inexplicable reason that hopefully does not spell out incompetence, Cyan continues to waste its reputation, and its fanbase’s time.