A Dashboard. Or a GUI Builder?

[jsakalos]

Hi community,

I'm thinking about creating a Dashboard extension. What do you think about it? Is there an interest? What features it should have?

[mdissel]

Can you show a screenshot to show exactly what you mean with a 'dashboard'?

[jsakalos]

No screenshot yet :(

Well, it would be a page space (a desktop) with pre-defined "landing-zones" (divs) in which you could put some content (with drag & drop for convenience). Then you could manage what is on the dashboard: reorder, remove, maybe resize if suitable.

The other "imagination viewpoint" could be like the dashboard is a blank papers page with some pre-defined spaces for various areas like columns, titles, pictures, etc.

Then you would be able to fill-in and organize the content of the areas.

There are many pages on net that use this concept but no link comes to my mind.

Could others help with a sample links?

[mdissel]

You mean something like http://www.igoogle.com, a portal with different sections containing all kind of data..

[jsakalos]

Yes, almost exactly that, Ext integrated with events, user proxy creation, D&D, maybe some integration with InfoPanel/Accrodion, etc.

[andytrickett]

Definitely interested - just looking at sticking a dashboard on the front of some of our stakeholder research/management applications. Might be exactly the kind of thing we're looking for..

[jsakalos]

Definitely interested - just looking at sticking a dashboard on the front of some of our stakeholder research/management applications. Might be exactly the kind of thing we're looking for..

Good!

Do you have any idea about features in (roughly) in the categories: must have, nice to have but can live without, nice but maybe never used?

You can setup your own categories, of course. This would influence the order of implementation if I decide to tackle it.

[brian.moeskau]

FYI, there will be a layout in 2.0 that does this. But don't let that stop you if you want it now :)

[jsakalos]

FYI, there will be a layout in 2.0 that does this. But don't let that stop you if you want it now :)

That was just an idea so I asked the community... ;) Anyway, thanks for info. Is there any link to a demo to see it?

[brian.moeskau]

We have a (somewhat) working version internally, but it still has some issues and has not yet been fully refactored into the 2.0 layout API. Once we have something worth showing, we'll post it (should not be TOO much longer).

[jsakalos]

Brian:

What will the dashboard layout in 2.0 be called? I looked around the pre-release build forum, but couldn't really find anything.

Jozef:

I want to be involved if you go down this route. I'm going to have to implement something like this, one way or the other. We're still in a pre-design / tools evaluation stage, so I don't know what I can do to help yet...

Nathan Beach

Thanks for giving hand Nathan,

I've looked to 2.0 sources too and I found ContainerLayout that could be near of the idea... I couldn't identify in the current code anything of the features I have in mind but you know, Jack is a wizard and he can show an example in 10 lines.

If Jack would provide a demo (together with a time frame of planned 2.0 release [+- several months would do]) or at least the planned features it would be much easier to decide if you even go into it.

[jsakalos]

We have a (somewhat) working version internally, but it still has some issues and has not yet been fully refactored into the 2.0 layout API. Once we have something worth showing, we'll post it (should not be TOO much longer).

Good Brian,

looking forward. :)

[nathanbeach]

Thanks, Saki...

Here's another good example (cross posted from Prerelease Builds forum):
http://www.dhtmlgoodies.com/scripts/...e-nodes-2.html

I'm still at the point of just trying to get my head around what all Ext can do, so I may (actually, will) be kind of useless for a while.

NBB

[jsakalos]

Thanks, Saki...

Here's another good example (cross posted from Prerelease Builds forum):
http://www.dhtmlgoodies.com/scripts/...e-nodes-2.html

I'm still at the point of just trying to get my head around what all Ext can do, so I may (actually, will) be kind of useless for a while.

Nathan

Hi Nathan,

(1) Ideas are never useless as far as they are constructive
(2) There are many ways to help (testing, commenting, solving cross-browser issues, donating, etc...)
(3) The link doesn't work (please edit, I'll look later)

Jozef

[nathanbeach]

Doh!

http://www.dhtmlgoodies.com/scripts/arrange-nodes-2/arrange-nodes-2.html

[jsakalos]

Do you think that (empty) proxy dragging would be enough? Like it's in the case of InfoPanel?

I've looked at the link and:

(1) drag performance is poor (probably caused by dragging all content)
(2) doesn't re-organize on the fly

My vision is to make it much, much, much better.

[nathanbeach]

I like "much, much, much better" much better. \:D/ I think that dhtmlgoodies link just reflects the basic functionality -- really iGoogle.com is a fantastic example.

Another interesting problem to solve would be layout "serializability". How do we make it so that the user can move his dashboard objects around and then come back, log in again (possibly on a different computer), and see the same layout?

NBB

[jsakalos]

Another interesting problem to solve would be layout "serializability". How do we make it so that the user can move his dashboard objects around and then come back, log in again (possibly on a different computer), and see the same layout?

Nathan

A serialized config object saved to server? Not sure if cookies would do with respect to size limits. BTW, what are default cookie size limits? Once I got some popup on over size but I just deleted 'em w/o further investigation.

[nathanbeach]

I.E. states a cookies size of 4096 bytes, but I don't know about F.F. Evidently Flash objects can store up to 100Kb...

I would prefer to avoid cookies because I wouldn't want the user's layout reset every time they clear out cookies, or access from a different computer. It may just need to be part of a bigger-picture user options / settings manager...

NBB

[jsakalos]

I.E. states a cookies size of 4096 bytes, but I don't know about F.F. Evidently Flash objects can store up to 100Kb...

I would prefer to avoid cookies because I wouldn't want the user's layout reset every time they clear out cookies, or access from a different computer. It may just need to be part of a bigger-picture user options / settings manager...

Nathan


It's not big problem to provide both methods cookies if user is not logged in (maybe we can set "forever" expiry on them) and server backed state if user IS logged in.

[aconran]

The Ext.state package would be the way to go for serializing the state of your objects when you login/logoff. So far there is only a CookieProvider but this could easily be extended to use a webservice which stored the state information in a database or use one of the propietary browser storage systems.

[efege]

Besides iGoogle, Netvibes is another example worth mentioning: http://www.netvibes.com/

[brian.moeskau]

In the version we're working on, the draggable portlets have a drop event that you can handle that gets passed information like column and index. With that, it's simple to persist your layout in whatever way you choose. Creating a custom state Provider (like CookieProvider) that can talk to your backend will probably be the way to go.

[jsakalos]

The Ext.state package would be the way to go for serializing the state of your objects when you login/logoff. So far there is only a CookieProvider but this could easily be extended to use a webservice which stored the state information in a database or use one of the propietary browser storage systems.

Yes, exactly this idea I have on mind.

[fredrik.stockel]

Imho Netvibes is absolutely a solution to be inspired by, Very slick!

[jsakalos]

In the version we're working on, the draggable portlets have a drop event that you can handle that gets passed information like column and index. With that, it's simple to persist your layout in whatever way you choose. Creating a custom state Provider (like CookieProvider) that can talk to your backend will probably be the way to go.

I was thinking about providing an API where user can register both sources and targets to allow initial dragging (like from tree and/or grid) and than methods of re-organizing 'em in the dashboard (layout).

There should be also overridable createGhost method and also createContent method if we're dragging let's say from tree.

As a summary it would be class' job to reorganize portlets but for initial creation the user methods would be called (maybe with some reasonable default methods to create at least something). Plus state management.

I also think that InfoPanels would play nicely with such a Dashboard (collapse/expand). Also (optionally) hidable Accordion could serve as a pool of panels/portlets.

[jsakalos]

Imho Netvibes is absolutely a solution to be inspired by, Very slick!

Agreed, very straightforward and intuitive. Only poor drag performance (at least on my 1.5GHz Centrino NB ).

[fredrik.stockel]

Are you planning a more or less complete dashboard API for both dashboard objects (portlets config) and the dashboard layout?

The InfoPanels can perhaps be reused to serve as a portlet-wrapper for almost any type of content/objects?

[jsakalos]

Are you planning a more or less complete dashboard API for both dashboard objects (portlets config) and the dashboard layout?

The InfoPanels can perhaps be reused to serve as a portlet-wrapper for almost any type of content/objects?

Sure, otherwise it doesn't make too much sense. These two classes have to play nice together.

Re InfoPanel: I was thinking reversed: the new DashbordItem (Portlet?) class would wrap InfoPanel, or any other (Ext) object and it would add API to work with Dashboard.

[franklt69]

Hi Saki I think is good to have Portlet in Ext, but if the team is working in it, maybe other thing will be more useful to Ext community, I don't know if in Ext2 will be :

1- tree with checkbox
2 - better PropertyGrid
3- Outlook-like calendar/scheduling control (http://www.daypilot.org/demo/Horizontal/)
4- Web calculator
5- other idea to complement some areas where Ext team don't work now
6....

I think it will do better Ext.


kind regards
Frank

[mdissel]

Hi Saki I think is good to have Portlet in Ext, but if the team is working in it, maybe other thing will be more useful to Ext community, I don't know if in Ext2 will be :


I agree.. and i add a Ext.ux.Wizard to list of possible extensions..

Thanks

Marco

[jsakalos]

I agree.. and i add a Ext.ux.Wizard to list of possible extensions..

Thanks

Marco

So good.

2 votes for 2 against.

What are we going to do?

[mdissel]

That's up to you, it's your time and effort for the community (a big thanks!). Although my preference is no dashboard (because it will be part of Ext). My vote goes to a solid / flexible Wizard (just like your Accordion Panel =D>) with Ext.Form support. (a couple of weeks ago i posted a first draft example in the Examples and Extras forum)

Thanks
Marco

[gkassyou]

perhaps, a Ext Gui builder to allow you to drag and drop components on a screen that you can then render the javascript from. I find alot of my time is spent on aligning the layout, creating forms, grid column models, and JSON readers.

Thx

[brian.moeskau]

In an effort to improve communication in this area, I've beefed up the User Extensions wiki page and also added a new Wish List section.

http://extjs.com/learn/Ext_Extensions

I think the best way to move forward on these types of ideas is for interested parties to adopt one of the current project pages (or create a new one) and start collaborating there so that the community knows what's going on. Note that I also linked from the wish list to the Ext road map forum thread (http://extjs.com/forum/showthread.php?t=4168) so that you can see what's already planned in Ext (the portlet idea has been on the 2.0 list for a while, it's just not finished yet).

If anyone has additional ideas for improving the process of promoting user extensions, please contribute! The discussion tab on the wiki pages is also a great place to post ideas or comments.

[Pagebaker]

I'm very interested in such a component, since I need it for a project I'm working on. I'm also willing to contribute:)
I don't have the time to wait on Ext2.0, so this is a perfect alternative!

[jsakalos]

I'm very interested in such a component, since I need it for a project I'm working on. I'm also willing to contribute:)
I don't have the time to wait on Ext2.0, so this is a perfect alternative!

Good Pagebaker!

Now it's 3 for 2 against.

Do you have and idea what features it should have? Best would be if you list your required features in http://extjs.com/learn/Ext_Extensions (http://extjs.com/forum/../learn/Ext_Extensions) Wish List and notify me in this thread - I read this more often.

[JorisA]

Hmm I'm voting against too.
Especially since this is a pretty big widget it would be a waste of time to code it twice. If your spending a lot of hours on contributing to the ext community I would vote for something new.

[TommyMaintz]

Vote against here aswell. Why would you want to make something jack is already working on. I love the quality of your work (accordion) and it would be such a waste to work on this.

[jsakalos]

OK,

3 for 4 against.

Two datums from Jack would help here:

1. Release date of Ext 2.0 +- several months
2. List of planned features of Ext 2.0 Dashboard (Layout)

You see, if Ext 2.0 is going to be released in 2-3 months then it's waste of work. If it's by the end of year, it's different. Together with the list of feature it makes easy to decide if to wait for Ext 2.0 and that's it, if to wait for Ext 2.0 and then extend a class or if to write it from scratch.

There is only one flaw connected with Ext and it's lack of any kind of schedule with planned features (and I don't believe that they don't exist internally in the Ext team). I don't know how you but I'd rather prefer known schedules with reasonable delays to being uninformed at all.

[gkassyou]

Jozef,

I still think a GUI builder makes sense. Any other takers?

[jsakalos]

Jozef,

I still think a GUI builder makes sense. Any other takers?


Sure it makes!

The GUI builder should also include a code generator and the full featured application like this is not a matter of couple of weeks and one guy but it's quite huge project for a team with reasonable time schedule and support.

To tell you truth, I hadn't in mind such a thing when I was creating this thread. The features what I had in mind were roughly:



a portlet pool (something like Accordion (http://aariadne.com/accordion) is now)
a portlet desktop - Dashboard with definable/resizable columns
a possibility to create/destroy new tabs with another Dashboards
drag & drop reordering/resizing of portlets
a locking feature or design (draggable)/use (locked) modes
collapse/expand of portlets (something like InfoPanel is now)
show/hide of portlets
user definable create/destroy routines (with reasonable defaults)
user definable drag proxies (with reasonable defaults)

I didn't mean it as application design tool, sorry. However, having the above features it should be possible to build on it.

[brian.moeskau]

Josef,

We don't have an exact feature set listed because it's still being worked on. I already mentioned that we have a working version that's actively being included in 2.0, so why would the precise feature set determine whether you start writing a duplicate version? Why not wait until we release ours and then enhance it if it's missing something you want?


a portlet pool (something like Accordion is now) - There is a similar accordion layout in 2.0 as Jack has mentioned before
a portlet desktop - Dashboard with definable/resizable columns - Already done
a possibility to create/destroy new tabs with another Dashboards - A portal layout is currently a single page. Not sure yet if tabbing will be part of the component or if you would simply contain multiple portal layouts within an existing tab set. Either way, creating new tabs within a portal should be easy.
drag & drop reordering/resizing of portlets - DnD is done, resizing may not be included (it's not currently -- it works more like standard portals like Google where portlet sizing is static)
a locking feature or design (draggable)/use (locked) modes - Already done
collapse/expand of portlets (something like InfoPanel is now) - Done
show/hide of portlets - The hooks are there to do this, although if a panel is hidden, it's up to the user to determine the UI for redisplaying it
user definable create/destroy routines (with reasonable defaults) - Done
user definable drag proxies (with reasonable defaults) - Well, there is a drag proxy, but to be "user definable" the developer would have to override the proxy. This is true for all Panel-based components in 2.0


We don't have a specific release date because of the magnitude of the changes in 2.0, plus it is still not feature complete. I would say that, roughly, we are targeting end of summer (maybe 2-3 months) but that is still a very rough estimate. The closer we get, the better estimate we'll provide.

There's lots of stuff that people have requested that we are not working on, as I listed in the user extensions wiki page. It would be great if people wanted to prioritize those and devote some energy there.

[jsakalos]

Super Brian,

this is exactly information I, and probably everybody, need.

Thanks.

[DigitalSkyline]

Saki,
You know how much I'm looking forward to this feature... but if its only a couple months down the road I'm sure my control panel can wait... I have plenty to keep me occupied until then :)

I also am looking forward to a timeline/scheduler widget of some sort, I was looking at the MIT implementation but it doesn't perform all that great... and their bug tracker is full of problems. Also DHTMLgoodies.com has a dragndrop scheduler widget that works pretty well, however I would love to have an Ext friendly control of some sort... anyways...

Time/money permitting I would just implement this all myself, unfortunately, with all the things I'd like to do, creating UI elements is about the last on my list... its just not what I'm good at :)

I know soon enough I'm going to have to pay the piper and give back to this community --- I think Ext is going to make all of us look good in our creations, so giving back is the least I can do. For now all I can offer is my thanks... Ext has inspired me more then any other technology in quite a while.=D>

[twhite56]

Saki,
You know how much I'm looking forward to this feature... but if its only a couple months down the road I'm sure my control panel can wait... I have plenty to keep me occupied until then :)

I also am looking forward to a timeline/scheduler widget of some sort, I was looking at the MIT implementation but it doesn't perform all that great... and their bug tracker is full of problems. Also DHTMLgoodies.com has a dragndrop scheduler widget that works pretty well, however I would love to have an Ext friendly control of some sort... anyways...

Time/money permitting I would just implement this all myself, unfortunately, with all the things I'd like to do, creating UI elements is about the last on my list... its just not what I'm good at :)

I know soon enough I'm going to have to pay the piper and give back to this community --- I think Ext is going to make all of us look good in our creations, so giving back is the least I can do. For now all I can offer is my thanks... Ext has inspired me more then any other technology in quite a while.=D>

What MIT implementation were you looking at? I'm looking at working on this exact project shortly. Thanks :)

[brian.moeskau]

http://extjs.com/learn/Extension:Timeline

[Eric Suen]

Jozef,

I still think a GUI builder makes sense. Any other takers?

A GUI builder you means a GUI builder build on Ext or whatever? Spket IDE will including a GUI builder and a Theme builder in feature, currently it still in plan stage...

[Pagebaker]

Ok, looking on others arguments and the Ext2.0 progress. It would be pretty much a waste of time. I need to be a little patient for a few months :)

[KimH]

We don't have an exact feature set listed because it's still being worked on.
I'm sure we all fully understand this B)

I already mentioned that we have a working version that's actively being included in 2.0, so why would the precise feature set determine whether you start writing a duplicate version?

Why not wait until we release ours and then enhance it if it's missing something you want?
Because we don't know the timeframe for this ;)

We don't have a specific release date because of the magnitude of the changes in 2.0, plus it is still not feature complete. I would say that, roughly, we are targeting end of summer (maybe 2-3 months) but that is still a very rough estimate. The closer we get, the better estimate we'll provide.
THIS IS exactly what we have been screaming about for a loooong time =D> Please update this thread (http://extjs.com/forum/showthread.php?t=4168) with this "roughly timeframe"... B)

[glauff]

I also vote for a GUI Builder and against a dashboard.
It would be very interesting to share ideas/questions about how to implement this,
and I am willing to contribute because we are in need of such an extension/application.

So I'll make a start:

Is it even realizable ? I personally think that ext has all that is needed for such a project, but I think it will grow bigger than we think.

So what are you thinking about an Ext Application Builder or GUI Builder ?

Is it something worth considering or just a waste of time an work.
My opinion is that it will become the killer app for the framework.

Greetings

Gaetan

[efege]

And I vote for starting a new thread titled "Ext GUI Builder project" (or similar), since this one is still too dashboard-oriented :)

[perrich]

+1 for GUI Builder

[jsakalos]

Hi guys!

Sorry for not answering for some time; I had some server issues to solve with customer.

So, Dashboard is dead let live GUI Builder!

Let me express my opinion; my 2 cents to the discussion. I was thinking about it and I've once written GUI SQL builder so I (approximately) know what it is all about.

Let's imagine that we already have the easier part: a tool that allows us to take an Ext element from a pool and place it somewhere on a desktop. Then there are two main areas to solve:



How and where to save GUI data like positions and sizes of elements, their relations (if any) and their configurable visual appearances. There are two main options: directly to the generated code or to "a database" (it can be a DataStore or something).
Code generator, and this is the main thing because sooner or later will come a requirement of two way "generator". GUI -> code generator and code -> GUI reader. This also can be done but then "don't dare to touch the generated code by hand" or you don't read it back to GUI.

And even all this could be done with these points in mind:



Underlying library has to be mature and stable. Until the library changes as fast as Ext does the GUI team - even if GUI would be already completed - wouldn't be doing anything else but implementing changes.
The GUI App Builder is not step by step project where the team can release a part of work done. It's either complete or not. Imagine that you can save positions of elements but event handlers not yet.

So this is big project for well coordinated team that does nothing else but that on the top of the mature and stable library.

Hope you don't kill me for this but I think that it's too early for a GUI Builder.

Cheers,

[gkassyou]

I'm glad others support the gui builder idea and I fully agree that unless you have a stable release of the EXT framework, the maintenance/updates will be insane.

In the meantime, a general conceptual design of the GUI could be discussed and started as well as the code generator framework. Perhaps a prototype of a lite code generator that will generate a grid using the simple grid properties widget.

ideas?

[JorisA]

I'd love to see a timeline widget (http://extjs.com/learn/Extension:Timeline)

[gregkats]

I think a great widget would be a Visio style widget.
http://www.openjacob.org/draw2d.html

[DigitalSkyline]

While were on the subject of widgets I came across this sitemap generator which is pretty cool:
http://www.scottjehl.com/localProcess/StyleMapV2/styleMap.html

A sitemap is generated from a standard html unordered list ( ul/li )

[Dmondark]

Hey guys!



How and where to save GUI data like positions and sizes of elements, their relations (if any) and their configurable visual appearances. There are two main options: directly to the generated code or to "a database" (it can be a DataStore or something).
Code generator, and this is the main thing because sooner or later will come a requirement of two way "generator". GUI -> code generator and code -> GUI reader. This also can be done but then "don't dare to touch the generated code by hand" or you don't read it back to GUI.I am just a newbie "coder" if you will :), but I have a couple of suggestions here that may be useful somehow:
I think the fastest way to go with this one is to save layout as xml (or any other Datastore as you said). A user would design the layout of a form (for example) and hit some save action, the layout is saved. Generating the JS code at this point may result in slow "saving".
Of course realtime xml->JS parsing is not an option here, as it will take forever to generate medium/big designs. So I think code generation would be at its best as an on-demand action. For example, you generate the code after you are done with your layout(s).
We have a (somewhat) working version internally, but it still has some issues and has not yet been fully refactored into the 2.0 layout API. Once we have something worth showing, we'll post it (should not be TOO much longer).

Any available demo for this yet?

Cheers,
Ali

[FreeGroup]

Hi,

I have create a demo for the Draw2D integration with Ext.

http://www.openjacob.org/draw2d/single/YahooUI_integration/index.html


greetings

Andreas

[ryedin]

Hi everyone... I just want to chime in here and say that we have indeed been working on a GUI builder (browser based IDE), called WebWidgetry - and we are building our base UI component library on top of Ext. Jack and Brian have seen some of our initial demo stuff actually and we've been trying to figure out the best way to get going a little faster.

I'd like to extend an invite to you all to help us out, especially those of you in this thread that have expressed interest in collaborating on such a project.

The stuff we have right now that is "demo-able" is built using our pre-Ext UI components, but I would be more than happy to set up some timeslots to walk anyone interested through what we already have (including some minimal Ext-based bits that are starting to come together already)... and start talking about getting any of you into our development team who want to be.

If you check out this post from my blog: http://www.someelement.com/2007/06/from-libraries-to-platforms.html - it explains the vision in a high level fashion.

I'll subscribe to this thread to see if any of you are interested - or you can email me directly at ryan dot gahl at nthpenguin dot com.

[Horace]

Seems to me that more abstraction = less speed.

[xmatt1]

Oh Man if u can make a dashboar.. like MAC OS

[alex.tomes]

Hi ryedin, I'll be happy if you can show us more about your application, I am interested in seeing/testing/commenting your web widgetry :)

[ben.tremblay]

You mean something like http://www.igoogle.com, a portal with different sections containing all kind of data..This thread is only 7 pages long (only?) but it is soooo full of good stuff.

I'm wondering if anybody reacted to your igoogle paradigm; I'm seeing NetVibes (http://netvibes.com/) doing some fine stuff with their "Universal Widget API (http://dev.netvibes.com/doc/)" (Dunno much about UWA but NV is one web-app/desktop I use daily.) and, well dang it, I would very much like to stick to ExtJS.

BTW: I came across an interesting thread in SalesForce "IdeaExchange" ... "Alternative AJAX Libraries" (http://ideas.salesforce.com/article/show/98360/Alternative_Ajax_Libraries). It goes without saying that I plugged this site.

[franckxx]

Hi,

A Netvibes or Igoogle-like can be really GREAT !!!

+1

[ben.tremblay]

A Netvibes or Igoogle-like can be really GREAT !!!
+1
I suppose the place to start would be widgetize Jack's RSS Reader ...
... or a POP3 monitor. (Dunno why that function pleases me so much.)

[franckxx]

Hi Ben.Tremblay,

I suppose the place to start would be widgetize Jack's RSS Reader ...

Do you see this Great RSS/Atom reader from YTorres ? just look here :
http://extjs.com/forum/showthread.php?t=19742
You can test here :
http://extjs.keliglia.com/
It's just... Brilliant !, the better RSS viewer i've see =D>

byebye

[ben.tremblay]

Do you see this Great RSS/Atom reader from YTorres ? just look here :
http://extjs.com/forum/showthread.php?t=19742
You can test here : http://extjs.keliglia.com/Desk-kregator, yes?
Thanks for the pointer ... I've been out of the loop for a while ... a very handsome piece of work!
(I don't want to branch this thread so will discuss at 19742.)

thanks again!
--bentrem

p.s. anyone interested in this really should look / see how NetVibes (http://netvibes.com) lays things out ... a matter of personal preference, of course, but theirs is very intelligent, IMNSHO a great exemplar.

[jackcrystal]

do you guys have that developed yet?


We have a (somewhat) working version internally, but it still has some issues and has not yet been fully refactored into the 2.0 layout API. Once we have something worth showing, we'll post it (should not be TOO much longer).