The Project
A satellite monitoring system displays performance data for airline Wifi service in a table. The table rows correspond to satellite receivers mounted in airplanes, and the table columns correspond to performance metrics or time-series values for a given timeframe depending on the report type.
Performance metrics are typically rendered as a group of columns whose column names may be derived from the performance data itself. System operators can choose pre-configured reports or build custom reports by selecting performance metrics on-the-fly. Changes to the selected metrics are rendered in the table immediately.
The number of reports and their column configuration is essentially unbounded, with new reports coming online in response to customer requests and internal reporting needs.
My Role
Sr. Application Developer and UI Architect who designed and implemented the data grid used for displaying the dynamic report configurations.
Interesting Challenges
The primary challenge for the dynamic data grid was reacting to real-time changes in the number and specification of the table columns, especially when the column characteristics are derived from the data itself. I designed JSONSchema column and report "descriptors" that specify column characteristics, cell styles, click events, data binding and transformations, which columns make up a report, and the source(s) of the dataset. When a report is selected from the ACL-mediated choices in the left-nav, the report descriptor and its column descriptors are used to retrieve the dataset(s), re-write the display template, and bind the dataset to the display template.
Two additional challenges include a) obtaining durable references to data items across multiple, sparse arrays in the dataset, and b) coalescing multiple report datasources into the smallest number of data requests. In the first case, I use a curried 'getter' to capture the array reference at the time the dataset is analyzed, deferring retrieval until the specific row is rendered. Coalescing multiple data sources is achieved by debouncing the data requestor until report selections have subsided and all data sources are known and can be combined.
Results
The data descriptors and parsers that construct the UI were deployed for over 12 months, and have been used with both Polymer and React component libraries. Two dozen custom and preconfigured reports were specified over time for use across multiple customers.
The Project
A satellite monitoring system displays performance data for airline Wifi service in a table. The table rows correspond to satellite receivers mounted in airplanes, and the table columns correspond to performance metrics or time-series values for a given timeframe depending on the report type.
Performance metrics are typically rendered as a group of columns whose column names may be derived from the performance data itself. System operators can choose pre-configured reports or build custom reports by selecting performance metrics on-the-fly. Changes to the selected metrics are rendered in the table immediately.
The number of reports and their column configuration is essentially unbounded, with new reports coming online in response to customer requests and internal reporting needs.
My Role
Sr. Application Developer and UI Architect who designed and implemented the data grid used for displaying the dynamic report configurations.
Interesting Challenges
The primary challenge for the dynamic data grid was reacting to real-time changes in the number and specification of the table columns, especially when the column characteristics are derived from the data itself. I designed JSONSchema column and report "descriptors" that specify column characteristics, cell styles, click events, data binding and transformations, which columns make up a report, and the source(s) of the dataset. When a report is selected from the ACL-mediated choices in the left-nav, the report descriptor and its column descriptors are used to retrieve the dataset(s), re-write the display template, and bind the dataset to the display template.
Two additional challenges include a) obtaining durable references to data items across multiple, sparse arrays in the dataset, and b) coalescing multiple report datasources into the smallest number of data requests. In the first case, I use a curried 'getter' to capture the array reference at the time the dataset is analyzed, deferring retrieval until the specific row is rendered. Coalescing multiple data sources is achieved by debouncing the data requestor until report selections have subsided and all data sources are known and can be combined.
Results
The data descriptors and parsers that construct the UI were deployed for over 12 months, and have been used with both Polymer and React component libraries. Two dozen custom and preconfigured reports were specified over time for use across multiple customers.
Skills
Demos / Artifacts