Hi,
You may have noticed that the Gmail Burglar Alarm could have stopped working some days ago. To get the application back working, simply, delete the calendars corresponding to the application.
The new calendars would be created once the application is activated again.
This was because of some issues with the calendar and the token that is used to access the calendar data. The issue seemed to spring up because of the way non-primary calendars were handled at Google Calendar. Deleting the calendar resets the token and this new token is used to add events to Google Calendar.
On a side note, I am also working on converting the plain and simple tabular data into better visualizations using Google Visualization API.
YAHOO Mail : Data Leakage Prevention
YAHOO mail has launched the ability to add applications to the web interface. This post is about an application that was presented at the YAHOO Hackday, Bangalore. The original hack is detailed in Babu Srithar's blog.
As shown in the video, the mail can be dragged onto the application icon to redact information. The prototype is crude in the sense that it uses encryption to achieve redaction. It also requires the user to identify sensitive information that has to be redacted. A production ready implementation could simply use industry standard data classification toolkits like Tablus and use it for identifying information. Using roles assigned to users, the redaction server can also ensure that information displayed to different depends on the roles assigned to them.
There is still some usability issues with having to drag and drop the mail to the application, but I think that YAHOO would overcome this issue and grant applications more capabilities.
As shown in the video, the mail can be dragged onto the application icon to redact information. The prototype is crude in the sense that it uses encryption to achieve redaction. It also requires the user to identify sensitive information that has to be redacted. A production ready implementation could simply use industry standard data classification toolkits like Tablus and use it for identifying information. Using roles assigned to users, the redaction server can also ensure that information displayed to different depends on the roles assigned to them.
There is still some usability issues with having to drag and drop the mail to the application, but I think that YAHOO would overcome this issue and grant applications more capabilities.
Sneak-O-Scope : Release 4.0
Hi,
We just released a new release of Sneak-O-Scope with a cleaner database. Though there were not many cosmetic changes to the applications, the code base is a lot more managable now.
The first change was the addition of the app engine console to the backend. The console is a simple python shell that executes commands at the server. This was very useful to test the database and the limitations imposed by the Google App Engine.
The second most important change was that the enter and exit time columns in the database were converted to the python datetime format. It was a textProperty earlier and that led to invalid date and times.
Another change was the introduction of a table to separately store unknown visits. This was done as querying for unknown and known visits separately is not allowed. Since we sort on enterTime in descending order, we could not have an inequality for unknown visitor. Hence, we had to separate them out into two tables and query them separately. This also helped us to pick 100 known and 100 unknown visitors, populating the list at the UI with atleast some known visitors instead of all visitors being unknown.
We also used the Preferences table to aggregate data. The table now has a Text column that stores the JSON for total visitors, known visitors, etc. The field will later used to store browser and friend statistics also.
We also included the simplejson library to parse JSON to and from the requests. This is better than parsing the strings to get data out.
There are also a couple of migration scripts that are to port data from the database at sneak-o-scope.appspot.com to orkut-scope.appspot.com. This would be the vertical partioning of the database for users from orkut ,myspace, etc.
The IP address has now been put to use with a YAHOO pipe that displays the location, ISP, etc in an iFrame. Details about this on a later post.
Last but not the least, we also included a HTML page to indicate that the application is not installed properly and has to be moved above other applications. There is also inline text next to unknown visitors about Orkut's privacy policy that does not let us display details of visitors who do no have the application installed.
We just released a new release of Sneak-O-Scope with a cleaner database. Though there were not many cosmetic changes to the applications, the code base is a lot more managable now.
The first change was the addition of the app engine console to the backend. The console is a simple python shell that executes commands at the server. This was very useful to test the database and the limitations imposed by the Google App Engine.
The second most important change was that the enter and exit time columns in the database were converted to the python datetime format. It was a textProperty earlier and that led to invalid date and times.
Another change was the introduction of a table to separately store unknown visits. This was done as querying for unknown and known visits separately is not allowed. Since we sort on enterTime in descending order, we could not have an inequality for unknown visitor. Hence, we had to separate them out into two tables and query them separately. This also helped us to pick 100 known and 100 unknown visitors, populating the list at the UI with atleast some known visitors instead of all visitors being unknown.
We also used the Preferences table to aggregate data. The table now has a Text column that stores the JSON for total visitors, known visitors, etc. The field will later used to store browser and friend statistics also.
We also included the simplejson library to parse JSON to and from the requests. This is better than parsing the strings to get data out.
There are also a couple of migration scripts that are to port data from the database at sneak-o-scope.appspot.com to orkut-scope.appspot.com. This would be the vertical partioning of the database for users from orkut ,myspace, etc.
The IP address has now been put to use with a YAHOO pipe that displays the location, ISP, etc in an iFrame. Details about this on a later post.
Last but not the least, we also included a HTML page to indicate that the application is not installed properly and has to be moved above other applications. There is also inline text next to unknown visitors about Orkut's privacy policy that does not let us display details of visitors who do no have the application installed.
Moving Sneak-O-Scope above other applications
There were a lot of queries about the installation of Sneak-O-Scope and it not working properly. Almost all of them were because the application was installed but not expanded by default. The users would have to move the application above other application on their profile page. Here is a video of how it is to be done.
The accident
I had an accident yesterday and am totally immobile now. With just a laptop to keep me company, I created this graphic reproduction of how it happened.
This was done in swish.
This was done in swish.
Searching through my delicious bookmarks
Hi,
I had earlier written about a ubiquity command that allows bookmarking pages quickly into delicious. It had help from YAHOO pipes to generate tags automatically and submit to delicious using their API.
The only part missing was a neat way to search those bookmarks. Since the tags were auto generated, delicious did not do a good job to look at the page content. Also, I would depend on Google's search index looking at only these bookmarked pages. The Google Custom Search was the answer, and here is my custom search engine that does the job.
Creating the search engine was simple, all I had to do was to include my delicious feed into the makeAnnotations Url. This generates the annotations file automatically, updating it at intervals. We do not have to go through the skull drudgery of creating annotation files, including sites, etc.
The search seems to be useful and I created the A9 search XML and added it to my search bar using the Mycroft engine builder.
If you have interesting delicious feeds, do let me know so that i could add your sites also, above the default Google results returned. The only thing that the custom search page lacks is the Images, Videos tag for which I am writing a greasemonkey script.
I had earlier written about a ubiquity command that allows bookmarking pages quickly into delicious. It had help from YAHOO pipes to generate tags automatically and submit to delicious using their API.
The only part missing was a neat way to search those bookmarks. Since the tags were auto generated, delicious did not do a good job to look at the page content. Also, I would depend on Google's search index looking at only these bookmarked pages. The Google Custom Search was the answer, and here is my custom search engine that does the job.
Creating the search engine was simple, all I had to do was to include my delicious feed into the makeAnnotations Url. This generates the annotations file automatically, updating it at intervals. We do not have to go through the skull drudgery of creating annotation files, including sites, etc.
The search seems to be useful and I created the A9 search XML and added it to my search bar using the Mycroft engine builder.
If you have interesting delicious feeds, do let me know so that i could add your sites also, above the default Google results returned. The only thing that the custom search page lacks is the Images, Videos tag for which I am writing a greasemonkey script.