On 10/15/13 5:09 AM, Jonathan Bernwieser wrote:
Hi there,
I am currently doing my Bachelor thesis at TU Munich, at the Software Engineering chair of Prof. Broy.
The goal of this thesis is to create a tool to automatically
categorize source code in open source software. Different categories
will be “test code”, “generated code” and “productive code” to better
evaluate and use the results of quality-check techniques. (Static
analyses might detect certain quality problems even though they’re not
relevant for a certain code category. One example would be the amount
of clones found in a project. It has to be checked what kind of
category the evaluated code belongs to as clones aren’t causing
quality issues if they occur in “generated code”.)
In order to create and test heuristics to identify code categories, I
first need to create manually a collection of different projects (or
classes to be more specific) I actually know about what kind of
category they belong to.
While manually going through the derby project I found generated files
in the following directories:
·derby\derby-10.6.1.0\src\db-derby-10.6.1.0-src\java\engine\org\apache\derby\impl\sql\compile\
·derby\derby-10.6.1.0\src\db-derby-10.6.1.0-src\java\tools\org\apache\derby\impl\tools\ij\
Are there any other generated classes I didn’t recognize?
Thanks you for your help.
Looking forward to hearing from you,
Regards,
Jonathan
Hi Jonathan,
The term "generated code" is overloaded with many meanings. This term
can be applied to several kinds of code associated with Derby:
1) The code generated into the "generated" subdirectory when Derby is
built. This code includes
a) The code generated by javacc from Derby's SQL dialect (found in
generated/java/org/apache/derby/impl/sql/compile)
b) The code generated by javacc for Derby's ij tool (found in
generated/java/org/apache/derby/impl/tools/ij)
c) Some shell scripts copied to generated/bin
d) A list of class sizes used to estimate memory usage at run time:
generated/java/org/apache/derby/iapi/services/cache/ClassSizeCatalogImpl.java
e) The English message source generated by
org.apache.derbyBuild.MessageBuilder from
java/engine/org/apache/derby/loc/messages.xml. The English message
source is generated into
generated/java/org/apache/derby/loc/messages_en.properties
f) Debug information generated into
generated/java/org/apache/derby/shared/common/sanity.
g) A demo database generated into generated/toursdb
2) Optimized chunks of localized message files are generated by the
localeinfo target of the Derby build into files which have names of the
form
classes/org/apache/derby/loc/m${partitionNumber}_${languageCode}_${countryCode}.properties
3) Compiled SQL queries, which include generated Java byte code. The
byte code is generated into memory in the org.apache.derby.exe package
space. This byte code is generated by Derby in production. In some
cases, this code is stored in an invisible column of SYS.SYSSTATEMENTS.
4) SQL text which is generated by Derby in production for certain
operations like bulk import.
The generated code you have found corresponds to items (1a) and (1b) above.
Hope this helps,
-Rick
