[madPL] Today’s Seminar Cancelled

[LAUREN MARIE NEUDORF <lneudorf@xxxxxxxx>]

Hello everyone,

Hope you all had a great break! Unfortunately, our seminar for today is being rescheduled, however, there are several great presentations being given through CS703 this week. The first two are today at 2:30pm, Engineering Hall room 2317.  Group 1 includes Cong Ding, and Austen Fan presenting on "Detecting Logic Bugs in Database Engines via Subset Mutation"

The second presentation, from Group 2, 

Venkata Saikiranpatnaik Balivada will be on "Generating jq parse expressions for JSON"
The abstracts for both are below. There will be more presentations on Wednesday and Friday, I will be sending information on those shortly. Hope to you there. 

Abstract for Group 1:

Software testing, as an effective approach to assure software quality, has been an indispensable step in modern software development. Database systems, i.e., Database Management Systems (DBMSs), are one of the most widely used large-scale software systems. Database developers often devote a great amount of time and effort to testing database systems to assure software quality. Software test engineers also develop some effective fuzzers for database system testing.

We define logic bugs as bugs that cause a query to return an incorrect result, e.g., by erroneously omitting a row, without crashing the database system. Logic bugs are highly hidden from the database system users, who cannot directly sense or realize this kind of bugs. Database developers also need to design specific test oracles to detect logic bugs. Researchers have proposed a typical test oracle for logic bugs in database systems. In particular, seed queries are mutated without changing the result set, and then the two result sets are compared to determine whether the database system under test includes logic bugs.

To tackle this issue, we design a new test oracle---subset oracle. In particular, a seed query is mutated while ensuring that the result set of the mutated query is a subset of that of the seed query. The mutation space is significantly enlarged and logic bugs are triggered with higher possibility.

Group 2's Abstract:

JSON is a ubiquitous data exchange format. For example: it is widely used in web API responses and pipelining processes in different scripting languages. We consider the second case: in our experience, sysadmins need to create a lot of quick and dirty scripts by clobbering together different programs. (for e.g. cat events.log | grep -i “’event: ‘LOG’” | ./filterToday | ./write). As requirements change, they need to come up with new scripts. New files might consist of a slightly modified structure, or the expected output might need to be stored in a different schema. Instead of constantly changing programs to match the expected input/output format, they prefer to modify the piped data stream on the fly. jq is one such widely used tool that allows transforming JSON data. Learning jq however, requires learning its non trivial DSL language. We build a solver that synthesis a jq DSL query to transform the JSON stream based on example data points.

Get Outlook for iOS