Update: Vik Singh had the idea for BOSS, and posted Yahoo! Boss – An Insider’s View. It’s money line is this, and describes the big idea succinctly: “I think users should be confident that if they searched in a search box on any page in the whole wide web that they’ll get results that are just as good as Yahoo/Google and only better.”

First, here’s what happened tonight:

Yahoo! is taking a bold step tonight: opening up its index and search engine to any outside developers who want to incorporate Yahoo! Search’s content and functionality into search engines on their own sites. The company that sees just over 20% of the searches performed each day believes that the new program, called BOSS (Build Your Own Search Service), could create a cadre of small search engines that in aggregate will outstrip their own market share and leave Google with less than 50% of the search market.

Might this impact things? He thinks so:

In both cases, Yahoo! BOSS is intended to level the playing field and blow the Big 3 wide open. We agree that it’s very exciting to imagine thousands of new Yahoo! powered niche search engines proliferating. Could Yahoo! plus the respective strengths and communities of all these new players challenge Google? We think they could.

And that part that was music to my ears (emphasis mine):

It is clear, though, that BOSS falls well within the companies overall technical strategy of openness. When it comes to web standards, openness and support for the ecosystem of innovation – there may be no other major vendor online that is as strong as Yahoo! is today. These are times of openness, where some believe that no single vendor’s technology and genius alone can match the creativity of an empowered open market of developers. Yahoo! is positioning itself as leaders of this movement.

Marshall, thanks for the great writeup. Yahoo!, thanks for making me proud.

everything in this article is my paraphrasing of speakers’ presentations. not my own words.

(Video coming soon.)

1. We run web applications. We’re only focused on this narrow goal.
2. We handle the entire lifecycle of an app.
3. Apps are run on Google infrastructure.

“It’s hard, but it’s worth it for us.”

“For the first time you can use the same infra we use…Auth, GOS, BigTable”

The Stack

1. Scalable serving infra
2. python runtime
3. SDK
4. Web based admin console
5. DataStore

Demo: App from scratch in 8 minutes.

More details

1. Scalable Serving Infrastructure: fault tolerant (redundant). Fluid: don’t need to schedule needs up front… more servers come online dynamically.
2. Python Runtime and Libraries. All tools are generic, so new languages can be dropped in later. Python used in same python available otherwise. Goal: you can use any language eventually. We don’t want to limit you.
3. SDK: Environment to develop apps locally. Avail for Linux, Mac, Windows today. (But can probably work anywhere.)
4. Admin Console: web-based admin console. (Looks like google finance meets google analytics.) Tools for request logs. Data explorer. Usage/quote numbers. App-version balancing. Can hook up domain (don’t need to run at *.appspot.com).
5. Scalable Datastore. Schemaless object store. Not a clustered sql thing. Instead based on BigTable. (Whitepapers online.) Horizontally scalable. Reacts to hotspots. BigTable instead of SQL is a big change, and may take some time to get used to. But we think you’ll come to like it. Schemaless means you can add a new datatype or entity whenever – no need to update your schema.

Now we’re looking at a Datastore Model Class.

GQL Query example

SELECT *
FROM Story
WHERE title = 'App Engine Launch'
AND author = :current_user
AND rating >= 10
ORDER BY rating, created DESC

Other Notes

Mail Sending API

no setup needed.

Make HTTP Requests

Authenticate with Google Accounts

Frameworks

The whole Django framework.

Guido van Rossum: Creator of Python and member of Google App Engine team

My passion is making life easier for developers. With python i’ve done that for decades. Now i’ve joined GAE team. Excited by potential. (and that python was first picked)

First time that GOogle has let third-party people run software on their infra. That’s fundamentally a big deal.

8:13 PM “We’re offing 100% of the python lang.”

8:14 PM – we don’t offer threads, but you won’t been it because of our scalable arch.

GAE uses a quota system so nobody monopolizes the infra.

me: if it’s so scalable, why do they need the quotes?

What’s Next?

• large upload/download support
• purchase additional capacity
• other language support
• offline processing.

Startup Camp is an unconference-style event that’s dedicated to bringing together the various members of the startup community for a face-to-face collaborative meetup where its the attendees that drive the agenda (in true unconference fashion).

I’m really looking forward to tasting the excitement in air and seeing all the cool projects. 100s of people have registered – it should be fun. (But the real reason work’s giving me the day to attend is so I can be on hand to help people realize their dreams using YUI.)

If you’re there, please come find me and say Hi (even if you don’t need YUI support).

Perhaps the so-called “social web” isn’t about connecting people (not about helping people socialize), but about information conservation: If a person chooses to do something — no matter how small — it’s inherently interesting, precious, and valuable. We’ve barely started to figure out what to do with this second-generation information. Where we have it’s been exciting, useful, and successful: Flickr’s Interestingness and Clusters, the notion of “watching” on Upcoming, the newer “people who looked at this ultimately bought that” in Amazon, and of course Google’s PageRank. The idea isn’t new, but it’s still under appreciated.

Here’s the paragraph from Wired that surrounds the words used in the video:

And who will write the software that makes this contraption useful and productive? We will. In fact, we’re already doing it, each of us, every day. When we post and then tag pictures on the community photo album Flickr, we are teaching the Machine to give names to images. The thickening links between caption and picture form a neural net that can learn. Think of the 100 billion times per day humans click on a Web page as a way of teaching the Machine what we think is important. Each time we forge a link between words, we teach it an idea. Wikipedia encourages its citizen authors to link each fact in an article to a reference citation. Over time, a Wikipedia article becomes totally underlined in blue as ideas are cross-referenced. That massive cross-referencing is how brains think and remember. It is how neural nets answer questions. It is how our global skin of neurons will adapt autonomously and acquire a higher level of knowledge.

Here’s the video, which was created by Michael Wesch, Assistant Professor of Cultural Anthropology at Kansas State University:

(via via)

I wasn’t able to find release notes anywhere, so it will take some time to dig around and see what’s new. (Granted, I didn’t look that hard – but where’s the readme?) There were just a few clues on their blog announcement, but otherwise just grab the download.

What’s new in YUI v0.12? Thanks for asking:

1. Matt Sweeney has contributed a potent new control, TabView, built with the same high-quality thinking obvious in his Dom and Animation utilities. Want to progressively enhance existing markup with useful but unobtrusive JavaScript? Us too. Prefer completely built-from-script controls? No problem. Want the tabs on the top, right, bottom, or left? All supported out of the box. You can populate the tabs with static on-the-page content, or, of course, pull it down on-demand with Ajax. It’s all good.
2. Adam Moore has completely reworked our generated-docs API documentation system (see the API docs for Dom), and it’s pretty damn slick. It’s much smarter now, and provides richer information cross-linked in more usable ways. Don’t miss the autocomplete-powered search on the API Docs main page. I was happy to read Carson’s comment on the YUIBlog: “[the] new documentation about brought a tear to my eye.”
3. Steven Peterson revisited his Calendar control in a serious way, and the results are great. In addition to the new and improved multi-calendar interface, he created in-depth tutorial-style examples of YUI Calendar highlighting all the key features and use cases for Calendar (as well as for the entire Container family). There has been more than one question on the ydn-javascript mailing list about how to do this or that with Calendar of Container, and he’s taken many of those and answers them definitively in the new well-written tutorials.
4. Eric Miraglia did selfless work, as always, to offer some key new features on the YUI site. Don’t miss the YUI Theater, with its ever-growing collection of video lectures and instruction (including great content from Yahoo!’s Douglas Crockford, and Firebug’s Joe Hewitt). On the home page itself, notice the piped-in live content from the mailing list and our blog; I hope that will bring even more people into the conversation. On each component’s landing page, notice one-click access to all the examples from the right column under the component’s cheat sheet. Eric has also brought all the cheat sheets up-to-date to this release; there’s a new cheat sheets for YUI’s CSS foundation files (Reset, Fonts, Grids), and for TabView.
5. The rest of the team has been busy too. Our director, Thomas Sha, improved Connection so that when you’re uploading files via setForm() and the asyncRequest includes a POST data argument, the appendPostData() method will automagically create hidden input fields for each postData label/value and append each field to the form object. Niiice. Jenny Han modified AutoComplete so that it’s a bit more efficient (always-on container don’t send show and hide events), and a bit more powerful (minQueryLength now supports zero and negative numbers). If the zillion options weren’t enough before, now you’ve got a zillion plus two. Todd Kloots didn’t rest either, and Menu now has more elegant internals, and a bit more functionality exposed.
6. For my part, I completed a pretty substantial rewiring of YUI Grids. The most exciting change is that Grids now offers Liquid/Fluid Layouts out of the box. At what cost? Just seven-tenths of a kb of new page weight. In addition, there’s more power, more stability, and more flexibility across the board. I’m a big fan of fluid layouts, but if Fluid isn’t your thing this release also has 950px page widths baked in, in addition to the original 750px width. Best of all, if you don’t want to use Fluid or the two preset sizes, it’s super easy to set your own custom width. The Template Presets and Nesting Grids offer the same functionality as always, but they’re a bit more bulletproof now, and they now enjoy spreading their wings within the new page widths. As before, the entire system is in ems and percents, so it breathes with the user’s font size – a favorite accessibility and usability feature of mine. The new system is fully backward-compatible, so give it a shot and let me know how it goes.

I hope you enjoy all the new features in this release. I’d love to hear your feedback in the comments below, or straight on the ydn-javascript mailing list.

Thanks,
Nate

I know, I know, you LOVE to go to work/school on a Friday. It’s your favorite day of the week and there’s nowhere you’d rather be than in your office/cube/classroom. That’s cool — I don’t judge — but, but, but next Friday (Sept 29th) you realllly should play hooky and sneak down to Yahoo for our first every public Hack Day and Developer Day. It’s gonna be quite the event, and I wouldn’t want you to miss it. Really, you should come.

Sold? Cool: learn more and request an invitation.

Developer Day, Friday from 9-5, is packed with 20 sessions across four tracks. They are not to be missed: Rasmus Lerdorf, the creator of PHP will be giving a talk. So is Iain Lamb, (an Ajax/DHTML pioneer who co-founded Oddpost which evolved into the new Yahoo! Mail product). JavaScript guru Douglas Crockford will be dropping knowledge, as will many others. Web-celeb and Flickr’s chief software architect Cal Henderson will be speaking. The plenary is by none other than Yahoo!’s VP of Product Strategy, Bradley Horowitz, (a very rare opportunity to hear him speak [for free]). Most of the authors of the open-source YUI Library will be guilding hands-on sessions dedicated to many YUI components (I’ll be teaching a hands-on session about YUI’s three CSS components.

But that’s not even the cool stuff!

We’ve got top-tier entertainment lined up for Friday night, and while they won’t tell me who it is, Michale Arrington (who’s in the know, and MCing this event) writes:

The entertainment lined up for Friday night is going to be incredible, although a non disclosure agreement prohibits us from saying who it is. I can say with confidence, though, that everyone attending the event will be very, very happy they were there for Friday night’s party.

Then the REAL fun starts: 24 hour hackathon.

Be there! (Tons of press will be, so if you miss it you can read or hear about it the next day.) We’ve got people coming in from Australia and everywhere between here and there, so beg borrow and steal and get yourself here too (You MUST register in advance – security will be tight… for real.)

Drop me a line if you want more info or whatever. Tons of people have blogged about this, so instead of linking to ‘em all I’ll just point you to Technorati: http://technorati.com/search/hackday.org?sort=authority

To these “services”, a license that deals almost exclusively with installed software doesn’t mean much. The software distributed under these various open-source licenses isn’t obsolete (in fact, I work on some fulltime), but rather these installation-based licenses aren’t sufficient or appropriate when “software as services” are concerned.

Granting somebody the rights to modify the source code behind the Yahoo! Term Extractor web service doesn’t make any sense. Instead, we need a way to license the service: How much capacity is provided? How much uptime is granted? What types of uses are legit? Etc.

This question that he’s raising makes good sense to me. I’ve got friends at agencies and startups that I encourage to use our extensive web service offerings. They want to (and do), but they have legitimate and real questions that a discussion like the one Tim’s provoking could begin to answer.

[1] I gotta get better about not losing things in the draft folder.

The 0.11 release last month brought with it the Logger Control and a host of other improvements to the library, including dramatically improved performance in the Drag and Drop Utility, file upload in Connection Manager, and color animations in the Animation Utility.

Beyond 0.11, the roadmap continues to hold to the course we’ve published in earlier updates. The best of our current thinking with respect to the next two release windows is digested below. The pipeline continues to include the Tab Control, the History Utility, and the Button control, all scheduled for the 0.12 release. For releases beyond 0.12, we have some early explorations underway; of these, the project we’re committed to getting on the roadmap is a table control, something we regard as crucial to any complete library and something we’re excited to add to YUI.

Next two release windows for YUI Library Beta:

1. August 21 (v. 0.11.3) — this will be a bug-fix update, addressing 0.11-release bugs in a variety of components.
2. Early October (v. 0.12)

Projects in Developmen

1. Tab Control

   The Tab Control will provide support for a variety of tabbed-module implementations.

   Projected Release: 0.12

   Confidence: High

2. Button Control

   The Button Control will enable the deployment buttons with (1) diverse visual treatments (e.g., with or without images); (2) configurable actions (clicking can be tied to form submission or other custom functions); (3) integrated menus and submenus.

   Projected Release: 0.12

   Confidence: Medium

3. History Utility

   Managing the browser’s history stack is critical to the creation of applications that are intuitive, usable, and sharable. Currently, management of the History stack in applications based on YUI requires you to roll your own solution. The History Utility will help facilitate this process by providing a simple interface for adding application states to the History stack during asynchronous interaction flows

   Projected Release: After 0.12

   Confidence: We continue to investigate actively the best approach to this problem across the A-Grade. We are pushing this back beyond 0.12 at this point based on what we’ve learned so far.

4. Table Control

   Dynamic tabular data controls are a common interactive treatment for data-intensive interfaces, going beyond simple table functionality to add features like dynamic sorting, editing-in-place, resizable columns, and more

   Projected Release: After 0.12

   Confidence: Medium

Note: This roadmap projects our plans over the next quarter or so; in so doing, it makes assumptions about conditions that are naturally dynamic. Some of the projects detailed here may be delivered earlier or later than we are currently expecting; some may not be delivered at all. Other projects not listed here may be escalated during this period. Use this document only as a rough guide; never rely on unreleased code listed here for any crucial needs.

Regards,
Eric

______________________________________________
Eric Miraglia
Yahoo! Presentation Platform Engineering