The goal was that each analysis and...

The goal was that each analysis and display tool to be included in the prototype analyzer should be designed and written simply once. Therefore, the data library is designed to normalize the variety of basic logic analyzer data marks and the variety of postacquisition data impressed signs generated by various analysis tools and to not absent this data to other analysis and display tools in a standard format.

Early in the design of the HP 16505A prototype analyzer we realized we requireed a new library for storing and retrieving logic analyzer data if we were going to engage our project goals. We destitutioned to duplicate the results of centurys of software engineer-years of effort in the HP 1650A/B logic analysis order and measurement modules. We also had to lay the groundwork to encounter the requirements of our prototype analyzer vision.

Understandably, we did not want to allocate enough commonalty to meet the first challenge for the duration of our relatively short schedule. However, looking at where the original time was worn out led to some insights. In the original HP 16500A/B combination of parts to form a whole effort had to be duplicated for each logic analyzer plug-in card introduced. That is, for each card, the lister, waveform, chart, and other software module povertyed to be rewritten. Some digest leveraging took place, but we did not flow close to complete code reuse. The different ways in which the analyzer cards at hand data to the software (largely because of differing memory layouts) had a great deal to do with this lack of reuse. without of this observation came the design goal that each analysis and display tool to be included in the prototype analyzer should be designed and written simply once. This meant that our library wanted to handle the variety of basic logic analyzer data symbols as well as accommodate the variety of postacquisition data adumbrations that our envisioned analysis tools would generate, and not past nor future this data to other analysis and display tools in a normalized format. Fig. 1 point out tos the reduction in the dimensions of the coding effort we springed to attain because of this library. It also point outs that we would have automatic commonality of feature puts across acquisition modules. Previously, a certain quantity of features for a new logic analyzer card were not quick in emergencies at the initial release of that card.

Several other design goals emerg that would allow us to come together our challenges:

* Retrieval time for analysis and display tools to access the data should be minimized. In the HP 16500A/B environment, the acquired data is examined in typically united display per acquisition. In the prototype analyzer environment, the goal was to encourage data exploration and to view and analyze acquisitions with several tools. We wanted to permit users to examine data simultaneously with the same view in different locations and with different views in the same location. These use gauges led to multiple accesses of the same data, in contrast with the HP 16500A/B example of a single view of data in a single location. Fig. 2 point outs the equivalent prototype analyzer graph of an HP 16500A/B use example while Fig. 3 shows a prototype analyzer graph with the multipleview, multiple-location use model

* Storage space for acquired data should be minimized. Given the potential size of a certain logic analyzer acquisitions (> 40M bytes) excessive expansion of the data was unacceptable. As always, however, there exists the trade-off between minimized retrieval time and minimized storage space.

* The storage mechanisms used from the library should be completely decoupl from application or client module That is, changing the way data is stored should not affect any lines of client digest Examples of client code include the tools that analyze and display the acquisition data of the like kind as the lister, waveform, chart, distribution, and pattern filter tools.

* The interface to the library should be a "natural" C++-style application programming interface. No just discovered style of programming should be introduced that would be different from normal C++ syntax.

* The library should provide a layer of memory management through the whole extent of the acquired data. Again, the possibility of highly large data sets makes it vitally important not to leak these enormous memory chunks or provide stale pointer's to nonexistent data. This goal is made more difficult according to the potential complexity of graphs possible in the prototype analyzer. Fig. 3 exhibits that there may be many analysis and display tools examining and viewing the data simultaneously. All of these regards to the data must be handled fitly to prevent large memory leaks.

* Finally, time correlation between different stations of data or analyzer acquisitions must be possible. This requirement makes it possible for users to examine their regularitys on several different hierarchical of the same heights concurrently, from looking at the analog nature of a country pin bouncing to looking at the high-level source digest executing at that moment in time. Time correlation of measurements also makes it feasible to intrigue one derived measurement against another derived measurement. Previously nonobvious relationships between variables may appear as a issue of these charts. Simply having the ability to impel markers in one view of an acquisition and watch that location automatically tracked in a different view of the same acquisition is another benefit of time correlation.