Some questions about your tool
taotao
I am a 2nd year master student, from Shanghai Jiao Tong University. I
also do some research work on static analysis. As we all know, the
existing static analysis tool, especially bug-finding tool, always
reports a lot of false positives. I want to do some work on about how
to improve the existing static tool's performance by ranking the
reported warnings.
Since iBug tool could extract the real bugs from bug tracking system,
I am wondering whether you could do me a research favor to provide a
prototype or binary tool. If that, I will save much time for my
further work; otherwise, I must manually confirm whether the reported
warning is a real bug or false positive.
Thank you very much. Any information about how to extract the known
bugs from history automatically is welcome.
Best,
Haihao
Valentin Dallmeier
Thank you for your interest in iBUGS. Our tool automatically links bug
reports to changes in the source code based on commit-messages and
other information. However, this still requires some manual
verification, as the data typically also contains quite some noise. I
would suggest you use one of the iBUGS subjects available at
http://www.ibugs.org . They contain bug reports together with
information where a bug was fixed.
Let me know if that helps!
Valentin
taotao
Thank you for your quick response.
First of all, I have to say that the link "http://www.ibugs.org" is
unavailable in China.
As described in the iBUGS webpage (http://www.st.cs.uni-saarland.de/
ibugs/), there is only one subject, namely AspectJ. In our research,
we would like to investigate more widely-used open-source projects to
validate the effectiveness of our error ranking approach. Could you
do me a favor to list all the subjects used in iBUGS?
Thank you again!
Best,
Haihao
On Apr 14, 1:36 pm, Valentin Dallmeier <valentin.dallme...@gmail.com>
wrote:
>
> Thank you for your interest in iBUGS. Our tool automatically links bug
> reports to changes in the source code based on commit-messages and
> other information. However, this still requires some manual
> verification, as the data typically also contains quite some noise. I
>
> Let me know if that helps!
>
> Valentin
>
Valentin Dallmeier
> First of all, I have to say that the link "http://www.ibugs.org" is
> unavailable in China.
http://www.ibugs.org is simply a link to
http://www.st.cs.uni-saarland.de/ibugs/ . Currently, there is AspectJ
and Rhino. Could you download the AspectJ subject and take a look at
repository.xml to see if it has the information you need?
Valentin
taotao
Thank you for your reply. Basically, I have known what you said. I
have another question:
Does need I download the all versions (Release 1.0, Release 1.1 ...)
of AspectJ dataset? Initially, I still see the detailed result for
AspectJ (a table with bugid, classname, and line number). But
currently, I could not see that again. Could you please provide the
link to it?
If I would like to locate the errors in other projects, could I reuse
the binary tool you provided in the existing two datasets?
Thank you very much.
Best,
Haihao
On Apr 14, 2:04 pm, Valentin Dallmeier <valentin.dallme...@gmail.com>
wrote:
>
> > First of all, I have to say that the link "http://www.ibugs.org" is
> > unavailable in China.
>
Valentin Dallmeier
> Does need I download the all versions (Release 1.0, Release 1.1 ...)
> of AspectJ dataset? Initially, I still see the detailed result for
> AspectJ (a table with bugid, classname, and line number). But
> currently, I could not see that again. Could you please provide the
> link to it?
You only need to download the last revision. You will find information
about where a bug is fixed in the file repository.xml . Each bug
contains a tag "<fixedFiles>", with diffs for all files that were
touched in the transaction that fixed the bug. Please also take a look
at the step-by-step guide at the end of our technical report
http://pages.cpsc.ucalgary.ca/~zimmerth/publications/files/dallmeier-tr-2007.pdf
>
> If I would like to locate the errors in other projects, could I reuse
> the binary tool you provided in the existing two datasets?
The tool is useless to you unless you have both the complete version
history and a dump of the JIRA bug database at hand for the projects
you want to analyze. iBUGS is focused on providing ready-to-use
datasets that other researchers can use to compare approaches. It's
not so much about the extraction process itself.
Regards,
Valentin
taotao
Basically, I have known what you said in the reply.
With the latest version as you meant, we could reproduce the detailed
bugs as described in Defect 2008. Right?
Another question is also the application on other projects. Generally,
for a well-maintained open-source project, the complete list of
versions is stored in their version server. Moreover, they also employ
the bug tracking system, like Bugzilla or JIRA. In light of what you
said, is there any possible to detect bugs for other projects with a
complete list of version and JIRA bug dtabase, when we adapt the
configuration on the existing datasets?
Thank you again!
Best regards,
Haihao
On Apr 14, 2:40 pm, Valentin Dallmeier <valentin.dallme...@gmail.com>
wrote:
>
> > Does need I download the all versions (Release 1.0, Release 1.1 ...)
> > of AspectJ dataset? Initially, I still see the detailed result for
> > AspectJ (a table with bugid, classname, and line number). But
> > currently, I could not see that again. Could you please provide the
> > link to it?
>
> You only need to download the last revision. You will find information
> about where a bug is fixed in the file repository.xml . Each bug
> contains a tag "<fixedFiles>", with diffs for all files that were
> touched in the transaction that fixed the bug. Please also take a look
Valentin Dallmeier
Sorry for the delay, I'm awfully busy at the moment.
> Another question is also the application on other projects. Generally,
> for a well-maintained open-source project, the complete list of
> versions is stored in their version server. Moreover, they also employ
> the bug tracking system, like Bugzilla or JIRA. In light of what you
> said, is there any possible to detect bugs for other projects with a
> complete list of version and JIRA bug dtabase, when we adapt the
> configuration on the existing datasets?
Yes, that's totally possible. Right now the tool works on subversion
repositories and JIRA databases. If you have more questions, please
contact me at dall...@cs.uni-saarland.de
Regards,
Valentin