Note: This is only summery
Automatically Documenting Program
Changes
What are motivations for this work?
Log messages are mostly with source
code. These messages are important component of software maintenance. The coder
can get some help by understanding editing, point and triage the defects. The
technical problem is that this log
documentation is burden to create and it may be partially complete or
inaccurate.
What is the work's evaluation of the proposed solution?
We introduce an automatic technique for
manufacturing concise, human readable documentation for arbitrary program
differences. For code summarization, our algorithm is based along the
combination of symbolic execution and novel base approach. The papers produced
by algorithm describes the result of a change on the run time behavior of the
program, it also includes conditions under which program behavior changes and
what the new behavior is.
What is your analysis of the
identified problem, idea and evaluation?
Mostly
developer spend their most of the time trying to read code. I guess it is good
algorithm that describe the consequence of a change of behavior of the program.
We discover that our generated documentation is suitable for replacing of existing
log messages that directly identify a code modification.
What are the contributions?
The principal one of the
contribution is an empirical, mathematical study of the use of the version
control log messages in many open source software organizations. Work shows us
there are many messages that that comprised with what and why documentation
also find that use is commonplace. An algorithm (DeltaDoc) is utilized for
identifying the varieties and condition under which they are occurring, combined
with a set of conversion heuristics the change summarization. By combing, these
techniques automatically generate a human readable description of code
modifications.
For objectively quantifying and
comparing the data capacity of program documentation a novel process is
applied. We try out this algorithm on a paradigm and a conflict of its yield to
250 human written messages from five projects. Our experiments supported by a
human study, which suggest DeltaDoc could replace over 89 percent of human code
generated what log messages.
What are future directions for this
research?
In future we can enhance our
DeltaDoc program efficiency by adding more techniques, adding more projects,
doing more brief experiments. We can increase the human written log messages
and by applying efficient algorithm the productivity is also increased.
Including the condition under which the program behavior changes and what the
new behavior is.
What questions are you left with?
I guess the main inquiry is that,
is this algorithm operates on a distributed network system with wide date. Is
the error percentage is more serious with another system and documentation
errors are minimized by adding some fresh techniques and experiments.
What is your take-away message from this paper?
We purpose a DeltaDoc, an algorithm
for fetching human readable code. Our technique is made up with symbolic
execution and a novel base approach to code summarization. It states us what a
code change affects. Our documentation
describes the result of modification of conduct of a program and what the new
conduct was.
