Актуальные научные исследования в современном мире iscience. In. Ua

«MODERN SCIENTIFIC CHALLENGES AND TRENDS»
 
SCIENCECENTRUM.PL 
ISSUE 3 
ISBN 978-83-949403-3-1 
36
SECTION: INFORMATION AND 
COMMUNICATION TECHNOLOGIES 
Chirikhin Aleksandr, Popova Kristina 
Novosibirsk State University 
(Novosibirsk, Russian Federation) 
 
 
METHODS OF DECREASING RUNTIME ERROR DIAGNOSTICAL MESSAGES 
REDUNDANCY 
Abstract
. Report redundancy problem is described and demonstrated. 
Reasons of report redundancy are identified. Algorithms reducing report 
redundancy are developed. 
Keywords
: runtime errors, C++, test, Valgrind, LLVM, memory leaks, 
redundancy. 
Introduction. 
It is common to divide errors that can be faced by a 
programmer during the development process into two parts: compile-time errors 
and runtime errors. If a program contains at least one error from the first part, it 
cannot be compiled. Syntax error is an example of such type of errors. If a program 
contains an error from the second part, a developer cannot be notified about it 
during compile-time [1]. Nevertheless, there are special tools which are intended to 
detect runtime errors during runtime. The most popular ones are compiler-rt, 
Valgrind and Dr. Memory [2]. Unfortunately, their diagnostical messages have some 
serious disadvantages making error analysis more complicated. Thus, if a program 
has one function and a thousand tests using this function, AddressSanitizer, which 
is a module of compiler-rt project, print a thousand of error reports.
The purpose of the work is a development of methods decreasing runtime 
error diagnostical messages redundancy.
Problem demonstration. 
Consider the following situation. There are three 
functions: doSomething, a body of which contains a runtime error, and test1 and 
test2 using doSomething. In case of isolated test running any runtime error analysis 
tool detects three errors although there is just one error in fact. Fixing doSomething 
function automatically fixes the other two errors so there is not any reason to spend 
time to analyze all these errors.