stealthtech's dashboard

After doing some further investigating I think the two *look* similar with a similar UX, but are two different technologies. Watching via Fiddler there are no similarities in how the data is filtered or retrieved.

Would this be something like a tool tip? If I hover over ERA I would get information or the acronym spelt out like Earned Run Average and that might even be a hyper link to Wikipedia on ERA? To your point that the XML needs to be changed to accommodate this along with changing Pivot. I would suggest adding an attribute called tooltip to the XML element FacetCategory.

"First off, keys are gone! Everyone can install without delay!"

"Pivot is currently available for anyone who would like to participate in this experiment with us." get pivot

It is now open to the public, that is just awsome!

How about Silverlight?

Thank you, I just had to make sure I avoided posting something about tiffany lamps. That Morton guy gets around, bubbles to memory storage, to tiffany lamps. I'm familiar with DeepZoomTools, but not so much with the other tools listed on the page. I use DZT for the Pivot Collection Maker, but it appears that the dynamic collection concept builds portions of a DZC where as the DZT builds the whole thing.

I like to have more questions than answers and to share my thoughts. I figure if I’m asking myself these questions then someone else out there has to be asking the same thing. I pose the questions not as to get an answer but to stir up the collective mind a bit. I like to have more questions than answers and I like to work with people that don’t have all of the answers. At times I might even have an answer (or think I do) but still ask the questions.

I read a sign once similar to this "!snoitseuq bmud on era erehT"

Thank you that was along the lines I was thinking. You can notice the UI/UX by looking at the Team which the link has one aspect of going to another collection and the Player going to a Web Page. But nothing distinguishing between the two links.

Also I noticed one drawback to this implementation. How do I go back to the Team Collection from the Player Collection intuitively? I can hit back several times until I get there. Depending on how deep into the Player collection I went this can be many steps? I can put a link on the Info Panel that would get them back, but not that intuitive. The mouse wheel, if I’m at the min zoom level and can zoom no more, but I continue to scroll towards me then the Team collection pops in? Is there an Up arrow, maybe next to the Zoom Slider Control? This would then associate the scrolling of leaving the collection which is similar to the slider changing when you scroll. Is there a way to know if the user opened the collection from another collection. If so then enable the button and allow scrolling out beyond the min level, otherwise keep it disabled.

On another note I tried the greater than less than and these two do go together. I figured this might be the case based on the GUI and your response. Because the UI has slider bars that give a range and not leaving one end open. Using the GE and LE to also include the value itself. for example:
Weight=GE.237.855670103093_LE.291.835051546392
I'm guessing if your range is 1 through 10 and you wanted everything greater than 5 you will need to do value=GE.5_LE.10

Here is an example using Link Facet type. Once you zoom to a team there will be a link in the info panel to Wikipedia.

http://pivot.forebrain.net/NFLTeams1/...

I would restate my previous questions to more of guideline as to the purpose and use of the Href and a Link Facet. Would one put the link to a drill down collection on the Href or as Link? As a suggestion maybe changing the glyphs or the UX/UI of a link to distinguish between links to web pages and staying within the collections. Similar to the same way an image is placed next to a link on a on a web page to designate the link as being an external link and not a link to a page within the website.

I just realized how deep I'm into Pivot that I'm at the point of not seeing the trees from the forest. There is a link facet type, but then the data would be Team Players for all items. Testing.

I get the concept with the business model and sectors. What I want to use is an everyday example by using the NFL Player collection with a Top level images of the NFL Teams.

The idea would be to have a Team collection which is one layer above all the players. That collection would have team collection with data points related to the team with maybe some rollup from the team players like salary. Once I click on a Team I would drill down to the Player Collection filtered on the Team selected from the Team Collection.

http://pivot.forebrain.net/NFLTeams/N...

One suggestion/question now that I have gone through this exercise. Would it be better to have a specific linking for this? Or multiple links? Or have an option that a facet type be a link? The reason I ask is that the functionality of going to a web page about the team is not an options. The Href as a dual purposes and if it used for drilling down to another collection then it can't be used for linking to an external web page.

Granted pivot will generate/guess a link "Bing Search Results" and "Releated Searh".

Well noted I will take that into account when building the helper functionality. Maybe only allow GT, LT on numeric values and EQ on strings and remove the other options. I will play around with it, I do understand the issue that you point out.

I got a laugh on the Caveat. That could be applied to the whole Pivot project ;) hehe, no? I think I have already taken enough risks and sacrifice of my time with the tools I have already built. Even though they might need to be retooled later when Pivot is released or thrown away if it is dumped. I am motivated because there is always something to be learned regardless of the outcome.

I don't think adding one more risk will incur too much damage to my soul. Just might mean more tooling later when the real deal is released. I think I have alluded to this catch-22 in other post. I have read things to the nature of "I won't build unless Microsoft builds" because it may not become a release product. Microsoft may not resale it if there is no momentum behind it. Catch-22. So, I take the risk with the hopes that Microsoft will see that people want this and are willing to take the risks. I’m hoping that the actions of doing something will speak louder than no action at all. BTW When is the beta going to be release to the public ;)?

Matt and I are also working on a SQL2Pivot project that will help build a Collection on the fly from data in a database. For now though, Pivot Collection Maker has the ability to import and export CSV files. You will also have the ability to load other CXML files to see how they are built. I’m going to do code review on the CSV functionality to make sure the bugs are worked out. One caveat is having the option of importing data that may or may not have the base CXML data (Image Location, Name, Href, and Description). Another caveat might be mapping the columns from the database schema to the table in Pivot Collection Maker. I offload this work on my SQL export. I will put together a walk through on how to build a collection and import/export data. The concept behind the projects are twofold one is to build a SDK for Pivot (Pivot Collection Tools) and the other projects (Pivot Collection Maker, and soon SQL2Pivot) are proof of concepts of what the SDK can do. Once the SDK is feature complete (in a week) you might want to try it for you own solutions.

Logan thanks for the heads up on the linking. I'm now going to figure out something I can do with that. Maybe make an intuitive front end that builds the QueryString of the URI for the HREF column on the users’ behalf. Maybe ask CXML: companies.cxml, Facet Category: Sector, Operator: EQ, Value: Retail. Then build "companies.cxml#Sector=EQ.Retail%20Stuff". I'm off to do my research fiddler2 here I come or what was that CTRL-ALT-U. The EQ seems to be a defacto standard would it be correct to assume that there are others like NE, GT, GE, LT, LE?

I have created a "simulated" Real World example. :)

Here is a link to my write up on what I'm expecting to see and what really is being displayed. This will also shows that a second sort will not help because the first and last name is not Facet of the item.
"Real World Sorting"
http://pivot.forebrain.net/

Here is a link to the collection.
http://pivot.forebrain.net/Comedian/C...

I'm still a little troubled with Pivot doing a hidden sort by name. I know that "how" I stumble across this could be a reason to write this off as not an issue. Who is going to have numbers as names?

I’m scratching my head to come up with a real world example that would be perceived as an error, because of the sorting by name.

The one example that I can think of is if I want to store the full name of the person in the Name attribute of the item. I create the CXML so that the items are sorted by the last name.

The CXML would look something like this:

<Item Id="1" Img="#1" Name="John Backman" >

<Item Id="2" Img="#2" Name="Betty Decker" >

<Item Id="3" Img="#3" Name="Albert Patrick" >

<Item Id="4" Img="#4" Name="Bart Randal" >

The results in Pivot would be:

Albert Patrick
Bart Randal
Betty Decker
John Backman

When I would be expecting this

John Backman
Betty Decker
Albert Patrick
Bart Randal

Another example is that I might have a custom sort order for the items being looked at and sorting by their name is not the way the custom sort works.

In some cases having a secondary sort might fix the problem. The example with the name would work because I would then choose Last Name as my secondary sort, but what if last name is not a data point to choose from? With the custom sort example this would not work because the sorting logic is not as simple as being numerical or alphabetical or work from a single data point hence the reason for having a custom sort.

I don’t think a secondary sort will be the answer and in some cases might just add confusion. Why not add a third sort or a fourth :). It might be nice to but not a need to.

I think having this sort or a better term display order be based on the item's id attribute and not the name attribute would be a better fit. That was my natural approach only to find out it was not that intuitive.

Forgot the other possibility is that I'm using it wrong. Thank God it was something simple. I almost gave up. After Binging the web a bit I found validation that the Morton Logic was correct which meant I was doing something wrong. The image list was in sequential order when it needs to be in Morton layout column major. I need to run the GetTileIds before calling CreateTile.

Instead of
1, 2, 3, 4, 5, 6, 7, etc.

The images need to be in Morton layout column major.
0 2 8 10 1 3 9 11 4 6 12, etc...

Now I can start to think of a Tile Handler, completing all of the handlers for an end to end solution (Collection Handler, Deep Zoom Collection Handler, and now Tile Handler).

I have made some progress on proving if the TileBuilder logic works. I'm putting this one to reset for a while I'm stumped. I built a control collection to get a DZC image to compare against the output of TileBuilder. They are not the same. I don't know if the changes I made had an effect. These where cosmetic changes that had nothing to do with the Morton layout logic. I might just have to dive into the number theory and see if I can see anything there. I'm hoping it is the cosmetic changes or something I did to get TileBuilder to work.

This is Zoom Level 5 Row 0 Column 0

Here is my control Image



Here is the output from TileBuilder:

That is the direction I was going in as well. I was first going after the image names. Which proved nothing. I just now created a third example using the padding in the CXML file for the item names.

Thank you. I published my findings here.

http://pivot.forebrain.net/

I have done some experiments. I can't say if the file name is the issue and if a file name with a width of one compared to a width of two makes a difference with sorting. Is 8.jpg different than 08.jpg and when sorted with 10.jpg it is after and not before (1 < 8)?
Well I did just that, I made my collection using ##.jpg and a collection with a mix of #.jpg and ##.jpg. No difference.

Mixed #.jpg and ##.jpg
http://pivot.corporate.forebrain.net/...

Unified ##.jpg with 0 padding.
http://pivot.corporate.forebrain.net/...

On a side note I can see distinctively the Morton Layout in the DZC using these images for testing. I will be using these as a control for a TitleBuilder. One item down and one more shows up ;) the fun never stops. I will keep plowing away and we will see what else is unearthed. I'm still hoping it is something I'm over looking with either the way I built the images, the CXML, or something.

I'm in the process of publishing a web page and will be hosting the assets used to build these collections.

I'm trying to understand the question about metadata and can say that I did nothing with an image's metadata to my knowledge. I hope maybe explaining how I created the images will answer the question. What I did was create an image (Paint.net) filled in the background with a given color, added a number, and saved. I then replaced the number with the next number and then saved as number x. I will double triple check my work to see if there is anything that stands out. I will also publish the collection and assets on the web soon.