Skip to main content

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
Re: [recommenders-dev] Stacktrace parser/detector

Was looking for some inspiration before starting to design and found this book http://searchuserinterfaces.com/book/

I'll take a look at it.

2011/6/4 Paulo Sérgio Medeiros <pasemes@xxxxxxxxx>
Answers below.


>> I have created the repository. You find informations on how to access
>> it here: http://wiki.eclipse.org/Recommenders/Labs
>> Projectname is "stacktraces"
>>
>> I think pure html as prototype is a good idea. Depending on the
>> framework we will use, we can easily extract templates from that html.
>> Marcel and me are both phd students at Darmstadt University of
>> Technology and we spend the whole day on that project. Okay, that's
>> not right ;-) We have to do teaching too and students are around all
>> the time.

Good to know, I'll set it up. I'll try the Sacha's suggestion and use Pencil instead of html. By the way, I wasn't aware about the tool. Seems to be good.


> I just had a discussion with Johannes which steps we should address next and we started to write an initial backlog for the stacktraces project. If interested, the backlog is a shared Google Doc we use to track progress of several sub-projects. Yet, we mainly used it as a shared whiteboard but we may formalize things a bit more if others start reading and editing it. Information how to access the backlog can be found here (spot the "Stacktrace Search Engine" sheet):
>
> http://wiki.eclipse.org/Recommenders/Infrastructure#Backlog
>
> ATM it contains the first brainstorming and prioritization of several tasks we think should be done next. You (all) may check out the stacktrace backlog and comment on it's current tasks and priorities if you like.

Took a look at that, but noticed that I don't have write permission (don't know if that was on purpose). Almost everything that comes to my mind is there. I would add some items related to the "social" or "human iteration" aspects of the app. Most come from the Identifying Software Problems Using Symptoms paper. One could be the possibility to allow the user to indicate if a stack is manifested because of invalid data or because of wrong logic. Another one could be the possibility to allow the user to indicate if two stacks "are the same", i.e., if they are caused by the same defect on the source code or environment. Do these things make sense? Probably more will come with the UI sketch.

And last, just a small correction my name is Paulo, not Paolo. :-)


> Johannes will start working on the backlog top-down, i.e., he will come up with an updated data structure of representing stacktraces in Java and JSON. I guess, the result should be checked by you since you also invested a considerable amount of time in representing stacktraces in Scala. Afterwards he will start designing several APIs and server-side services. In the meanwhile, I will spent some time on crawling Eclipse forums to build an initial database for stacktraces just to see whether the concept of crawling forums works as we expect.
>
> Does this sound OK for you?

Sure!

One doubt though, I saw backlog item related to lucene "Simple stacktrace search w/ Lucene". Isn't that part already implemented? Also, you said that you are updating the data structure for representing stacktraces, what are the motivations for that? Can you describe more how are using Lucene and how you use that representation to feed Lucene to create its indexes? Maybe you want to send me some papers that you have written or the Johannes' thesis if this is easier.

One last that i'm pretty sure that you are aware (but it is always worth remembering) is that Java has its own stacktrace representation http://download.oracle.com/javase/6/docs/api/java/lang/StackTraceElement.html. It was from there that I took some ideas.


> Just to provide you with little more background:
>
> Johannes did his master thesis on mining stacktraces and already wrote large parts of the stacktrace search engine, parser, and evaluation logic. He was advised by Eric and me. The implementation we currently present at various Eclipse DemoCamps is almost completely written by him. Thus, he's the "implementation expert" :)
>
> I'm ATM serving as discussion partner most of the time. I'm currently finishing my PhD thesis, am involved in teaching, and am in the happy position to advise a great team of 15 students doing theses or hands-ons (have to say this - they are on the list :-)) jointly with Johannes. My major focus is currently on community building, improving the existing parts of Code Recommenders, and pushing towards the "IDE 2.0" vision.
>
> Thus, to answer your question regarding how much time we will spent with stacktraces: We'll spent around 1 day per person and week - at least for the moment. Depending on Johannes workload, he will spent more time on stacktraces if possible. If my writing up is done, things will likely go up to 2-3 days per week and person. I hope this answers your question  how much time we plan to spent with this and what experience we bring into the project.

I would like to provide you more about me too:

I study at Federal University of Rio de Janeiro. Some information regarding my research group can be found at http://lens-ese.cos.ufrj.br/ese/. My advisor is Guilherme Horta Travassos. My main research focus is Experimental Software Engineering. I don't have a fully defined PhD theme, but it will involve Software Engineering, Machine Learning and Experimental Software Engineering knowledge management. I almost took the stacktrace parser/searcher as my master's theme, but it was too unrelated with my group focus so I wasn't able to convince my advisor. But now with my PhD beginning and with its relation to machine learning and AI techniques in general (not sure what technique I'm going to use yet), I renovated my interest in exploring AI things and so I have stared implementing the stacktrace idea to keep myself sharped and prepare for future implementations in my phd.

Regards,
Paulo.


On Fri, Jun 3, 2011 at 7:07 PM, Marcel Bruch <bruch@xxxxxxxxxxxxxxxxxx> wrote:
Hi,

little off-topic but interesting:

the stacktrace search engine has been proposed as new feature at StackOverflow: http://meta.stackoverflow.com/questions/93653/stack-trace-wiki-on-so However, it seems that the idea wasn't clearly enough presented by the requestor. It already got 6 down votes. Anyway, I think the search engine is a great fit for StackOverflow. We'll see whether the mood changes over time :-)

Best,
Marcel

--
Eclipse Code Recommenders:
 w www.eclipse.org/recommenders
 tw www.twitter.com/marcelbruch

On 03.06.2011, at 17:17, Marcel Bruch wrote:

> == this one got a bit lengthy… sorry for that :) ==
>
> Hi,
>
> I just had a discussion with Johannes which steps we should address next and we started to write an initial backlog for the stacktraces project. If interested, the backlog is a shared Google Doc we use to track progress of several sub-projects. Yet, we mainly used it as a shared whiteboard but we may formalize things a bit more if others start reading and editing it. Information how to access the backlog can be found here (spot the "Stacktrace Search Engine" sheet):
>
> http://wiki.eclipse.org/Recommenders/Infrastructure#Backlog
>
> ATM it contains the first brainstorming and prioritization of several tasks we think should be done next. You (all) may check out the stacktrace backlog and comment on it's current tasks and priorities if you like.
>
>
>>> I would like to ask is how much time you are dedicating to the project.
>>> I'll try to ajust mine to your pace.
>
> Just to provide you with little more background:
>
> Johannes did his master thesis on mining stacktraces and already wrote large parts of the stacktrace search engine, parser, and evaluation logic. He was advised by Eric and me. The implementation we currently present at various Eclipse DemoCamps is almost completely written by him. Thus, he's the "implementation expert" :)
>
> I'm ATM serving as discussion partner most of the time. I'm currently finishing my PhD thesis, am involved in teaching, and am in the happy position to advise a great team of 15 students doing theses or hands-ons (have to say this - they are on the list :-)) jointly with Johannes. My major focus is currently on community building, improving the existing parts of Code Recommenders, and pushing towards the "IDE 2.0" vision.
>
> Thus, to answer your question regarding how much time we will spent with stacktraces: We'll spent around 1 day per person and week - at least for the moment. Depending on Johannes workload, he will spent more time on stacktraces if possible. If my writing up is done, things will likely go up to 2-3 days per week and person. I hope this answers your question  how much time we plan to spent with this and what experience we bring into the project.
>
>
> Regarding next steps:
>
> Johannes will start working on the backlog top-down, i.e., he will come up with an updated data structure of representing stacktraces in Java and JSON. I guess, the result should be checked by you since you also invested a considerable amount of time in representing stacktraces in Scala. Afterwards he will start designing several APIs and server-side services. In the meanwhile, I will spent some time on crawling Eclipse forums to build an initial database for stacktraces just to see whether the concept of crawling forums works as we expect.
>
> Does this sound OK for you?
>
> Best,
> Marcel
>
>
>
> On 03.06.2011, at 09:03, Johannes Lerch wrote:
>
>> I have created the repository. You find informations on how to access
>> it here: http://wiki.eclipse.org/Recommenders/Labs
>> Projectname is "stacktraces"
>>
>> I think pure html as prototype is a good idea. Depending on the
>> framework we will use, we can easily extract templates from that html.
>> Marcel and me are both phd students at Darmstadt University of
>> Technology and we spend the whole day on that project. Okay, that's
>> not right ;-) We have to do teaching too and students are around all
>> the time.
>>
>>
>> 2011/6/3 Paulo Sérgio Medeiros <pasemes@xxxxxxxxx>:
>>> Ok, I'll work on the UI. Just to let you know that I'm not a specialist on
>>> that, but I'll do my best. :-)
>>>
>>> I'll start with the web UI. I was thinking to use pure html to implement the
>>> UI prototype, do you suggest something different? By the way, I will take a
>>> look at http://grepcode.com/ as Sacha pointed.
>>>
>>> As soon as you have set up the repository, please let me know. Another thing
>>> that I would like to ask is how much time you are dedicating to the project.
>>> I'll try to ajust mine to your pace.
>>>
>>> One last question, you have already mentioned students and the university
>>> repository. Are you professors? Students? From which university?
>>>
>>> On Wed, Jun 1, 2011 at 3:44 AM, Marcel Bruch <bruch@xxxxxxxxxxxxxxxxxx>
>>> wrote:
>>>>
>>>> Hi Paulo,
>>>> thanks for the papers. I was aware of the latter two (Bettenburg and
>>>> Schröter) but didn't know the former two.
>>>>
>>>> Great idea to have a crawler in platforms like bugzilla, I didn't thought
>>>> about that. In my “roadmap” I was only thinking about users inserting the
>>>> stacks via IDEs or the web interface and then discussing about them using
>>>> the web interface.
>>>>
>>>> Yes, a web interface to add and search for stacktraces, an Eclipse UI that
>>>> finds, searches and collects Userfeedback, and a crawler that fills an
>>>> initial database to start and continuously updates with latest stacktraces
>>>> found in Issue trackers, forums, and mailing lists etc., and a couch db that
>>>> stores these stacktraces along with their information where they have been
>>>> found/discussed + groupings of related stacktraces. That's roughly the idea
>>>> of what we would like to see at the end.
>>>>
>>>> I'm using Scala (sbt + specs + intellij) and Mongodb. I was planning to
>>>> use Scalatra or Bowler for a rest interface to this core. I was using them
>>>> only to learn some new technology, so I'm don't have any problem to change
>>>> that.
>>>>
>>>> That's good to hear. I also would like to work with Scala but I'm not sure
>>>> whether is is the best way since it needs to be embedded into an existing
>>>> OSGI architecture on server side and Eclipse UI and Web UI and maven build
>>>> system. It would take some effort to get this running and I'm not sure
>>>> whether the integration of scala in the current situation is valuable.
>>>>
>>>> At the stage that you are I think that the web ui is natural step. In
>>>> fact, I was just finishing the basic matcher to try to sketch some UI. So,
>>>> for me it ok to go with that. Lets try that. I've idealized some screens in
>>>> my mind, but hasn't stop to draw anything. And, as you said, the UI is a
>>>> really important component of the system. It should turn the search natural
>>>> and should really help to discuss and find useful information about the
>>>> faults. Some touch of social web (or Web 2.0) can be thought also such as
>>>> users indicating that one stack is related to the same fault of another
>>>> stack. But this is only in future versions.
>>>>
>>>> Having some sketches would be great! In the meanwhile we have to set up a
>>>> GIT repository for this and move the parts for the server and Eclipse UI
>>>> into this repository. We will start out with a university repository. After
>>>> some progress we can ask the Eclipse PMC to get committer access for you on
>>>> Code Recommenders and move the repository contents to Eclipse.
>>>>
>>>> Finally, regarding the papers, well as a PhD student this is always
>>>> welcome :-). Nevertheless, I was really wanting to push that to a “real”
>>>> project and not wanting to stop only on the prototype. I don't know what are
>>>> your expectations about that.
>>>>
>>>> That's great to hear (especially the last part :) ). Since you are a PhD
>>>> student I just wanted to point out that joint research papers is also in the
>>>> scope of this project. However, what drives us is the vision and how
>>>> beneficial such a system might get. The efforts we put into code
>>>> recommenders is by far more than required for a pure research project. Thus:
>>>> We want a real project too :)
>>>>
>>>> For the evaluations, the Eclipse bugzilla dataset will be very useful. I
>>>> work as software architect at Petrobras and can conduct some evaluations
>>>> there in the future ;-). My research group expertise is Experimental
>>>> Software Engineering and my masters was about using Action Research
>>>> methodology in Software Engineering and we can think about future studies.
>>>>
>>>> Sounds good. Let's get back to this when the prototype is up an running.
>>>>
>>>> So, how to continue?
>>>> Paulo, you start with some Web UI sketches?
>>>> Johannes , can you create a separate repository for stacktraces?
>>>> @Johannes, @Eric
>>>> Since this project now involves people w/o committer access at Eclipse I
>>>> think we have to split the repository for this and create a new, public one
>>>> on vandyk, right?  If you agree we can start moving things into this new
>>>> repository one by one and integrate them into the build system.
>>>> Best,
>>>> Marcel
>>>>
>>>> _______________________________________________
>>>> recommenders-dev mailing list
>>>> recommenders-dev@xxxxxxxxxxx
>>>> http://dev.eclipse.org/mailman/listinfo/recommenders-dev
>>>>
>>>
>>>
>>> _______________________________________________
>>> recommenders-dev mailing list
>>> recommenders-dev@xxxxxxxxxxx
>>> http://dev.eclipse.org/mailman/listinfo/recommenders-dev
>>>
>>>
>> _______________________________________________
>> recommenders-dev mailing list
>> recommenders-dev@xxxxxxxxxxx
>> http://dev.eclipse.org/mailman/listinfo/recommenders-dev
>
> _______________________________________________
> recommenders-dev mailing list
> recommenders-dev@xxxxxxxxxxx
> http://dev.eclipse.org/mailman/listinfo/recommenders-dev

_______________________________________________
recommenders-dev mailing list
recommenders-dev@xxxxxxxxxxx
http://dev.eclipse.org/mailman/listinfo/recommenders-dev



Back to the top