Tuesday, May 4, 2010

Billboarding and Scripting: The Mysterious Spacer Image

A lot of progress has been made on the Kukui Kup Billboard and most of the functionality we set out to do has been implemented. We did get a little lost on the way with four people doing the same project and all, but we just had to keep reminding each other what things needed to be done and someone always stepped forward to do it. This is one huge advantage that I realized when working in a group where there seems to be more than enough people. When you finally identify some specific task that needs to be done, there will always someone willing to take on the job, or at least there will be someone to delegate the task to. ;)

When we decided on the layout we were going to use last week, we also needed to lay out the functionality we wanted the Billboard to have and we essentially came out with this list:
  • Apply more advanced CSS concepts
  • Transitions for modules
  • Create Google Visualization mock-ups
  • Energy Tips
Currently, the most up-to-date version features all of the above objectives, and can be found at our project hosting site.

For now here's one screen shot of the billboard in action:

There's been a lot more style applied to it from last week's mock-up. There has been more style applied to the billboard overall and especially the ticker on the bottom. We also incorporated a BioHeatMap Visualization, sadly the gadgetized BioHeatMap that I worked on previously is incompatible with web pages due to proprietary gadget methods being called. But since this is more of a read-only type of deal, it doesn't matter if the gadget was used or not.

Next, here's another screen shot of the billboard:

I got it mid way through a transition showing that it fades out to do so. It is also showing an energy tip in the middle.

My work throughout the project has mainly dealt with the ticker and the visualizations. One problem that I did come across with the ticker was that when it first started, it never started entering from the right.

Here's a screen shot of the desired entrance when the page first loads:
(Notice it entering from the right side)


Where as here's a screen shot of where it was actually starting out from:
(Started from the left side and started scrolling)


Reading the semi-instructions for the ticker, it seemed like I was doing nothing wrong. I mean, how would I error if I was just copying/pasting code. I finally decided to ask Kendyll to help me in identifying what was going on. So we sat down together and we both were staring at JavaScript for about an hour... What we came across was that in the source code, the JavaScript uses a spacer image that stretches the length of the ticker to make it appear that it is entering from the right. But the instructions just said to use the .js source code and that's all you need. But we needed to make our own spacer image in order to get the desired affect. I guess the lesson here is that you can't fully trust instructions from another developer, and that even the most simple of instructions might be too simple and you forget a step to include.

Tuesday, April 27, 2010

Synchronous Scriblink

Heading into this new endeavor for the Billboard Mockup has become to a slight stagnant halt due to the lack of cohesion and communication. In past projects such as WattDepot-Apps and the BioHeatMap, the group size has never exceeded 3 people, including myself. But it seems that four people working on a single task deserves much more attention to detail in delegating tasks. One of the main lackings we faced was the ability to communicate with one another. With half the group with different schedules it was difficult to find time to talk about the billboard.

Previous collaborations yielded simple text messengers or exchange of e-mails sufficient, however, the billboard requires communication that cannot be conveyed in the two former mediums. For this unique situation, I remember using a site called scriblink.com with a friend who needed help with his Physics homework. Essentially, Scriblink is free online JavaScript-based white board that supports multiple users. Upon entering Scriblink you are automatically put into a private white board channel, from there you can invite people to your white board and each person has the same privileges to write/draw/import pictures all at the same time.

Although Scriblink is most enjoyable with a tablet (where you can freely draw like on paper), all you need is decent mouse skills to draw whatever. There's even a chat box on the side so if you can't participate in the drawing, you can still communicate textually to everyone else.

With the use of our new found tool Scriblink, a meeting was arranged with Kendyll, Jarret, and me, where we mostly talked about the specifics on the layout for the billboard. Here's a screen shot the mock-up we currently have:


Whereas here's the final 2 layouts we worked out on Scriblink:

The specifics for the columns are to be shrunk since some elements such as the Floor Standings can be tightened and made room for prizes/ads.

We also reasoned that if we sorted out and delegated tasks between the two sub-groups (Me and Kendyll, Jarret and Paul) we would still be working with those who we are most comfortable with and still be on task. Each portion of the module will be split between the two groups. Since Kendyll and I have been working on Google Visualization's the most out of our Software Engineering trek, anything that deals with charts/tables will be handled by us, where as prizes/ads will be dealt by Jarret and Paul.

Tuesday, April 20, 2010

Kukui Kup Continuous Display: What to Display?

With the conclusion of the WattDepot BioHeatMap gadget, I recently became involved in creating a billboard display for the Kukui Kup Competition. Ideally what this entails is a display for when students enter the lobby of their dorm, they can see various aspects of the competition such as dorm rankings, recent commitments, and upcoming events instantly as they walk by. This idea of a lobby display would be a cool addition to continually remind students to save energy and go green. Robert did some price comparing and this Samsung 46" LCD display seems sufficient to be the actual medium to display the billboard. It also features a built-in Windows computer as it will use a web browser to display content.

Tag teaming with the other gadget group, Jarret Mizo and Paul Galiza, Kendyll and I will be working together to create a functional mock-up for the billboard. To maximize development without worrying about cross-browser issues, we decided to develop everything using only one browser, the latest version of Firefox (3.6.3). Jarret recommended this plug-in: https://addons.mozilla.org/en-US/firefox/addon/1568 to install which removes the tabs at the top in Firefox when you Full Screen giving Firefox a clean look, which is a nice feature adding to the overall awesomeness of our developing billboard.

We'll post all of our functional mock-ups at a newly created Project Hosting site here.

In order to get a feel for the billboard I took a look at the mock-ups that were already done by a previous group here. It's got some great examples, especially the ones with the "Did You Know?", they'd make great transitions between the different set-ups. By the end of this week, we will have a functional mock-up of the possibly looking like the following:


The frameworks that we'll be using is a combination of HTML, CSS, and JavaScript. Since WattDepot already speaks Google Visualization, we'll mostly likely be using the Table Visualization to create any of the table data. Prof. Johnson also recommended some kind of ticker at the bottom of the page showing recent updates. I've been searching for an RSS Ticker that integrates nicely and this one: http://www.mioplanet.com/rsc/newsticker_javascript.htm looks promising.

Some modules I've been tossing back and forth with Kendyll are:

Tuesday, April 13, 2010

Go Go "Prototyped" Gadget: BioHeatMap v1.0

The completion of the BioHeatMap Gadget outfitted for WattDepot sparked one last test before totally labeling it as "finished." This was the cross-browser test to make sure that it did work properly in other browsers. So far, much of the work has been done in the latest version of IE, v8. This is sort of a contradictory action due to my past work in WattDepot-Apps. Much of the work was done through Safari and Firefox. However, since this was my first time coding for a Google Gadget, I felt a little more inclined to attach "training-wheels" to the whole process and start with Google Gadget Editor. But to my dismay, Google Gadget Editor does not load properly running Mac OS X Snow Leopard, on Safari 4.0.5 and Firefox 3.5.9. So I hoped on over to my Boot Camp'd Windows 7, and in IE and Firefox, the Google Gadget Editor worked, so I ended up doing much of my work there.

When it came time to view the BioHeatMap in Safari and Firefox, the gadget wasn't resizing it's height properly, and it seemed as though it stayed at it's default height of 200 pixels at all times. Here's a screenshot of what the gadget looked like w/o resizing:

This was a complete surprise to me since it resized perfectly in IE. I spent a good portion of my weekend trying to figure out how/why it wasn't resizing. Of course, the method that's supposed to do the auto-resizing is "gadgets.window.adjustHeight();" The first thing that I checked was that it was being called properly, which in fact it was... I scoured the Gadget Help Group to see if any other people had problems with the adjusting the height. But what I found was references to Google's deprecated method "_IG_AdjustIFrame();" which did not help. I tried replacing "gadgets.window.adjustHeight()" with "_IG_AdjustIFrame()", manually adjusting the height with window.innerHeight/window.outerHeight, nothing I did seemed to remedy the situation.

Looking at other Google Visualization Gadgets that used "gadgets.window.adjustHeight()" those gadgets resized perfectly. So it got me wondering, what's different about the BioHeatMap than other Google Visualizations. While sifting through the code I remembered that the BioHeatMap wasn't a visualization made by Google, so it had to use the Prototype library, an external open source JavaScript framework, to load the BioHeatMap. The "prototype" gave me another keyword to use to search for a solution.

Finally, I came across an existing issue in another Google Hosting Project. The issue found here: http://code.google.com/p/opensocial-resources/issues/detail?id=104 described exactly what I've been looking for. So turns out, there wasn't anything wrong with the way I was calling "gadgets.window.adjustHeight()", but the actual Prototype library itself and how it interferes with adjustHeight. The actual specifics of the problem can also be explained in the issue. About half a days worth of working out this problem and searching for a solution can be summed up into 3 lines of code...


Along with the auto resizing, I also implemented onboard time and date pickers for the BioHeatMap that is placed directly on the gadget itself without using the Gadget's Edit Settings menu. It was a real pain in the butt to code these things since all the HTML markup had to be manually coded and set. But nonetheless, I like how it turned out:
(Above is a picture of the last 24 Hours of real data from Prof. Johnson's house in Kailua, which features a Time Picker)
(Above is a picture of the last 14 Days of real data from Prof. Johnson's house in Kailua, which features a Date and Time Picker)
(Above is a BioHeatMap of multiple sources of UH's own Saunders Hall, displaying Kendyll's implemented multi-source capability)

Overall, I'm happy with the results. Creating a Google Gadget was a neat experience in that the output can be readily shown and easily distributed across many individuals just by linking to an XML document. However, the actual coding of the gadget itself was a little gruesome. Perhaps I've been spoiled by Eclipse and other IDE's, since I was mostly bounded to Notepad for this gadget. But now, instead of looking at lines on a chart, or numbers in a table, we can look at colorful blocks on simulated or live energy immediately showing trends and times of less and deep usage of energy.

The Google Hosting for all the WattDepot Google Gadgets can be found here.

Tuesday, April 6, 2010

Flux Capacitor of Transposition

Our BioHeatMap Google Gadget is well underway and almost complete. We've managed to move away from using Google Spreadsheet as our data source and can now take in Data Source URL's from WattDepot and user specified source to output a BioHeatMap of either the last 24 hours, last 7 days, or last 14 days. These options were chosen because of the width constraints set by Google. We didn't want a gadget that required scrolling, but wanted it to give a general idea of energy usage..using colors instead of boring numbers!

The most up-to-date version of the Gadget can be found here.
Whereas here's a screen shot of it running on my iGoogle page:
(Using Firefox and showing Last 24 Hours of SIM_HPOWER)


I'm extremely pleased with the output of this Google Gadget. Needless to say, this scant 3 liner gadget took a while to become what it is. After getting the query URL setup, and finally getting the auto-size height working, Kendyll and I ran into a huge problem with the visualization. Our very first WattDepot BioHeatMap visualization came out looking like this:


This is a step backwards because our proposed view of the gadget was to have the columns as dates, and the rows as sources. What we got instead was a completely transposed version of what we intended. The first thing I did was go to the Google Visualization API and look up the methods for the response tables. But to no avail there wasn't anything about a "transpose" method. At that point, I sent Kendyll a message on my progress. Kendyll scoured the Visualization Group to see if anyone had done this before, but his efforts for fruitless. So Kendyll decided to take on the task of creating a new Data Table and manually copying values to transpose the response table.

In order to get an idea on how to transpose the table, we had to get a little more information on what a response data table looks like. Using the toJSON() method on a response table, this is what a response table from a query to WattDepot looks like:


The table is column-typed/based, meaning each column defines the type of data that's placed in each cell for it's column. The columns are defined first ('datetime', then 'number' as above), then populated by rows. So in our case we would want the first column to be defined as a 'string' to list all the sources, and the rest of the columns to be defined as 'numbers' which hold the numerical value of whatever data has been queried. For the most part, the heart of the transpose function was implemented by Kendyll, however, to cross the finish line, he needed a fresh pair of eyes to look over the code as there were some minor details he over-looked. In this case, the column labels are strongly typed, and need to be a string. So we were using Date objects and had to explicitly call the toString method in order for the Data Table to process correctly.

Here's a snippet of the transpose function: (It takes in a response data table and returns a new table that contains transposed values)


After some final formatting of the dates, since blatantly using the Date objects toString method is extremely long for a Google Gadget, I coded a method to take only the parts that we needed and the final output of the gadget currently looks like my inserted picture above.

I added issues to Issue Manager for our Google Project, but the next thing we need to do is implement multi-source capability, because right now it only takes in a single source, but ideally it will be comma separated.

Tuesday, March 30, 2010

Gadgetizing BioHeatMap: Colorful Squares of Energy

Shifting gears to coding a Google Gadget is a very interesting and different experience than previous endeavors. My past projects such as Greenometer and WattDepot-Applications have yielded fruitful results, but creating a Google Gadget is something that can be immediately distributed and displayed directly on the web without any complication between servers or multiple frameworks.

With my partner Kendyll, and using his idea of implementing a BioHeatMap visualization, we decided to take on this new challenge as one step further in creating a tool student's can use for the UH Energy Dorm Competition.

An example of a BioHeatMap visualization looks like this:
It essentially takes in a matrix of data and conditionally formats the data by displaying various colors. It finds the max and min values, assigns two colors to them and all other values in-between, assigns color appropriate to how close those values are to the max and min. We are currently exploring two options:
  1. Rows are the floors. There are 24 columns, one for each hour of the day. The cell is the *cumulative* energy consumed during the competition for that hour of the day. That gives people on the floor a sense for when they're using a lot, as well as how they compare to others.
  2. Rows are the floors. There are 28 columns, one for each day in the competition. The cell is the total energy consumed by each floor on that day. (A cell is blank if the day is in the future so there's no data for it yet.) This shows which floors used a lot on a particular day. If a floor is mostly red, it's using a lot of energy, and if a floor is mostly green, it's using very little. It also highlights which floors are consistent vs. which floors are "spiky"---one day using a lot of energy, the next day using very little.
Since this is something relatively new to take on, the first thing on the agenda was to get a working "dummy" version of the BioHeatMap visualization as a Gadget. By looking at the BioHeatMap visualization page, it was a little discouraging to see that most of the development for the BioHeatMap has ceased since the latter part of last year. But by looking at other visualizations that are backed by Google, I came across the Google Visualization Gadget Gallery. Sadly the BioHeatMap was not listed under here since it was originally created by Institute for System Biology, but I was still lucky enough to dissect one of the listed gadgets, mainly Gauge. I chose the Gauge since another group is going to try and implement it for the competition, so I helped myself to see what they would be dealing with.

The actual .xml file is being hosted in the Project Hosting here.
But for those who want to try out the "dummy" BioHeatMap the link to add to your iGoogle page is: http://wattdepot-ui-googlegadgets.googlecode.com/svn/trunk/BioHeatMap/test-bio-heat-map.xml

The data source is just a Google Spreadsheet: http://spreadsheets.google.com/ccc?key=0AhWy1L2qDnFYdGtuQTlJME9rN2pDOFhGRFBuMHZsbFE&hl=en.

Right now only I and Kendyll have access to edit the Spreadsheet, so if you want access, just send me an e-mail and I'll add you to the Google Doc.

Some of the issues I'm currently at is the default height of the Gadget is extremely long. I will have to correct that this week and try and get it to fit "snug" with the visualization. A general issue I came across is while browsing the example pages for the BioHeatMap. Most of the examples look pretty cool, however this one caught my eye:
It looks like a total mess, and judging by the size, I'd say that's a typical size we'll be looking at. Although I'm not sure how sporadic the data has to be in order to get it like that, but just by first glance it's hard to interpret or make heads/tails.

Tuesday, March 16, 2010

Advanced Visualizations, New Frontier on Familiar Grounds

As the planning of the UH Dorm Energy Competition continues, there are still other things we are considering to implement. Shifting from WattDepot Applications, I am going to be working on a new topic, Advanced Visualizations. This new group is sort of like the WattDepot-App's Visualizer, except we are going to explore the different Google Visualizations and see how we can apply some of them to WattDepot.

The main group consists of , but I will be working with Kendyll Doi, the same Kendyll Doi from WattDepot Applications, as a sub group.

In order to get started, Prof. Johnson provided a few suggestions of what we could do. Right now, I'm just looking at the different suggestions and getting the feel of how I would implement them. Details about which framework we should use, either Wicket or Google Gadget, will be sorted out sometime this week.

Here are some of the suggestions that I think would be duo-able:

Rolling Update of Sensor Data
This visualization basically combines the idea of the monitor (i.e. get the latest value) with the trend line (show changes over time).
The idea is to have a visualization where you specify the source(s),the update interval (i.e. 30 seconds, 1 minute, 1 hour, 1 day), and the window size (i.e. 5 minutes, 1 hour, 1 day, 1 week), and the result is a chart that is refreshed every update interval and shows the last window size of data.
The reason why I like this idea is that in theory it sounds like the WattDepot Visualizer, but with an added refresh event and without the messy-ness of multiple sources, although in theory it could accommodate multiples (since it already does that). The only problem I'm seeing right now is that if the refresh interval is set too short, say 30 seconds.

Gauge
An alternative visualization for the "monitor" gadget. Also has red/yellow/green capability like the stoplight.
A simple enough visualization, similar to a speedometer in a car. However, the challenge with this one is that the example Gauge visualization shows that the limits between red, yellow, and minor are hard-coded values out of 100. My thoughts would be to try and scale all the values to 100, but that would require some type of maximum calculation for all data being presented, and since the data would be streaming live, it would be hard to set a number to scale to 100.

Indeed there are a lot more other Visualizations to choose from, but at this point the refreshing annotated timeline, and gauge seem to be the most plausible. The other visualizations such as Bio Heat Map for displaying energy generated and Term Cloud for quickly displaying a source that's consuming the most energy also seem plausible, but Kendyll and I are going to play around with the refreshable timeline, and possible Gauge to get us going. Our first weekly meeting for this new group with Prof. Johnson is scheduled for this Thursday. More details to come as Kendyll and I decide which framework and visualization to attempt to implement.