SER 2009

Software Engineering 2.0 and Research 2.0

 

TTC 2010 on Twitter

Posted by: Pieter Van Gorp

Tagged in: Untagged 

Pieter Van Gorp

Thanks to the Epsilon and Spoofax people for Twittering during TTC:

  1. Image_normal richpaige RT @dskolovos: @louismrose wins the #ttc10 model migration case-study award with Epsilon Flock #tools10 #icmt10
  2. Image_normal richpaige RT @dskolovos: another award for @louismrose and Flock: best exogenous transformation in the live contest of #ttc10
  3. Computerscience2_normal UoY_CS RT @dskolovos: another award for @louismrose and Flock: best exogenous transformation in the live contest of #ttc10
  4. N222303216_3337484_6879_normal louismrose Tassilo Horn wins #ttc10 live contest with GReTL http://yfrog.com/j16k5xj Lost track of the other awards!
  5. Epsilonlogo_normal epsilonews RT @dskolovos: @louismrose wins the #ttc10 model migration case-study award with Epsilon Flock #tools10 #icmt10
  6. Me2_normal dskolovos another award for @louismrose and Flock: best exogenous transformation in the live contest of #ttc10
  7. Me2_normal dskolovos @louismrose wins the #ttc10 model migration case-study award with Epsilon Flock #tools10 #icmt10
  8. N222303216_3337484_6879_normal louismrose Jens von Pilgrim and Mitra win the Ecore to Genmodel Case in #ttc10 #tools. I think Fujaba won the Topology case (but not sure!)
  9. N222303216_3337484_6879_normal louismrose Triple-graph grammars make an appearance at #ttc10 #tools10 Hiding under the table and watching through my fingers :)
  10. N222303216_3337484_6879_normal louismrose Markus now presenting COPE's convergence view for reverse engineering of metamodel changes. Very neat #ttc10 #tools10
  11. N222303216_3337484_6879_normal louismrose Sebastian Buchwald presents GrGen.NET and Markus Herrmannsdörfer COPE http://yfrog.com/4b7u9oj http://yfrog.com/2t364vbj #ttc10 #tools10
  12. N222303216_3337484_6879_normal louismrose And now Rueben Jubeh presenting their migration solution in Fujaba #ttc10 #tools10 http://yfrog.com/0733uqj
  13. N222303216_3337484_6879_normal louismrose Agris Sostaks presenting MOLA for model migration at #ttc10 #tools10 http://yfrog.com/b84vbwj
  14. N222303216_3337484_6879_normal louismrose Tassilo Horn presents GReTL in the #ttc10 workshop at #tools10 http://yfrog.com/123mjtj
  15. N222303216_3337484_6879_normal louismrose Tassilo Horn presents GReTL in the #ttc10 workshop at #tools10
  16. Bernhard_merkle_small_normal bernhardmerkle @EelcoVisser cool can you add the eclipse project (for the lambda) to this blog ? :-) #tools10 #icmt10 #ttc10
  17. Iktwitter_normal zef RT @EelcoVisser: Now with screencast: "Lambdas in Spoofax" a blog post about my entry for TTC 2010 http://bit.ly/aMF660 #tools10 #icmt10 #ttc10
  18. Acherm_normal acherm RT @EelcoVisser: Now with screencast: "Lambdas in Spoofax" a blog post about my entry for TTC 2010 http://bit.ly/aMF660 #tools10 #icmt10 #ttc10
  19. Kirikaavatar48x48_normal ShinNoNoir RT @EelcoVisser: "Lambdas in Spoofax" a blog post about my entry for the live Transformation Tool Contest http://bit.ly/aMF660 #tools10 #icmt10 #ttc10
  20. Self_normal EelcoVisser Now with screencast: "Lambdas in Spoofax" a blog post about my entry for TTC 2010 http://bit.ly/aMF660 #tools10 #icmt10 #ttc10
  21. Image_normal richpaige RT @louismrose: Wrap up at #ttc10. Many implementations of lambda calculus DSL in less than 6 hours. Impressive! Tomorrow the results are compared #tools10
  22. Self_normal EelcoVisser "Lambdas in Spoofax" a blog post about my entry for the live Transformation Tool Contest http://bit.ly/aMF660 #tools10 #icmt10 #ttc10

Transformation Tool Contest 2010: Live Contest Challenge

Posted by: Pieter Van Gorp

Tagged in: Untagged 

Pieter Van Gorp

Steffen Mazanek is currently presenting the live contest challenge for TTC.  The domain under consideration is lambda calculus.  This may sound theoretical at first but the concrete challenges involve many practical model to model and model to text issues.  

The reference solution to the challenge is contained in this ZIP archive and can be evaluated in SHARE here.  Note that during the contest you don't really need the SHARE demo, since the zip archive contains sufficient input and ouput models (a test suite for all transformation solutions).

Have fun, be sharp!


TTC 2010 opening slides online

Posted by: Pieter Van Gorp

Tagged in: Untagged 

Pieter Van Gorp

The slides describing the workshop program, publication strategy, etc. are available at Slideboom.

Regards, and enjoy the workshop,
Pieter


I have finished a new screencast (PDF copy) as documentation for the new SHARE feature that enables people to stress-test each-others solutions using input models that they have stored locally.

Last year's tool contest indicated that many people required some mechanism to upload content to SHARE and at the time, we had to give people full internet access in the remote virtual machines... which is obviously undesirable on  a larger scale since:

  1. SHARE contains various artefacts that should not be downloadable to the local machines of SHARE users,
  2. one could abuse SHARE for hacking other sites or send SPAM etc. (although we could trace this back based on IP information it is better to prevent such activities.)

In a nutshell, SHARE users have now a private folder on the virtual machine server (currently we do not allow more than 500MB per user but we will revisit such restrictions on regular basis).  Users can use rsync to transfer data from their local machine to their remote SHARE folder.  This is for example useful when a reviewer wants to exercise the correctness of a research contribution further by using some of his own test inputs.

As a convenient side-effect of this new feature, users can now also store data across immutable sessions.  Previously, state transfer between sessions was only possible when using a mutable clone, but now the 500MB folder will persist across any session. 

Personally, I already find this very useful in the context of various teaching activities, since now TU/e students that use a Mac can work for example with CPNtools without requiring a mutable disk (which saves us a lot of hard-drive space).  Moreover, in a research context, it makes a lot of sense to have a private server folder in SHARE because that enables you to test for example all tool contest solutions to the BPMN2BPEL case study using the same input BPMN models.

Enjoy, and please do not hesitate to send me some feedback (via e-mail or via the discussion features of this site),

Pieter Van Gorp


Collaborative tool evaluation platform

Posted by: Pieter Van Gorp

Tagged in: Untagged 

Pieter Van Gorp

Hi all,

I really look forward to R2oSE 2010. 

As an online warm-up, I would like to get some community input for the following issue: during the yearly transformation tool contest [report], we typically have an interactive session to derive a feature matrix for comparing the solutions that were submitted to our case studies.  We start with some brainstorming, then cluster features on a blackboard, and finally make three feature matrices that we print on paper and hand out to all workshop participants.  This enables a live peer review: solution submitters get about 15 minutes to demonstrate their transformation program and the audience fills in one matrix column in the meanwhile.

By assigning weights to the different features, we then analyze which solution satisfies the criteria that were proposed by the audience the best.  The following table aggregates the data from  9 evaluations of 10 contributions: BPMN2BPEL solution evaluation results.

The advantage of this interactive approach is that the winner of our contest is usually supported by most of our workshop attendees.  Currently though the manual process of having the brainstorm session, clustering on the blackboard and then creating an spreadsheet is very time-consuming and the final step (creating the actual spreadsheet, documenting the meaning of a criterium, ...) needs to be done by one of the organizers (or we simply run out of time for considering the actual solutions...) This is undesirable, since it allows (probably accidental) bias and it is too error-prone.

Therefore, I have been thinking of web 2.0 support for streamlining this process.  The idea is to involve workshop participants as soon as possible in the creation of the digital evaluation form (or underlying feature model) and to enable some online conflict resolution/integration.  I am currently looking for a website that provides:

  1. an account system (since we want to avoid anonymous rant) [example: planet-research20.org, but several alternatives are out there],
  2. a (semi-)structured editor for feature matrices [could be google spreadsheets, a feature diagram editor, ...]
  3. (optional) a mechanism for merging contributed feature matrices (reconciling different documents/models created using the editor from (2),
  4. a mechanism to assign weights to features (to enable the automatic computation of the "best" solution,
  5. (optional) a mechanism to override the weights (on a user basis) after the workshop.  That would enable end-users to assign a winner according to their own preferences.  Usually, they would do that for finding which solution they could use out of the box or which solution they would like to extend, or ... (whatever we currently cannot imagine yet)

I can imagine this might at first sound rather specific to our transformation tool contest (TTC workshop) but I think that such a platform would be very useful for various other research purposes (researchers often want to structure a discussion about which algorithm/tool/... performs "best" according to too implicit criteria!) Once we better understand how to use (& improve) such a platform, we have a new means to perform survey research: instead of doing a large literature review, installing some tools, creating a feature model and assessing the result based on our personal understanding, we have a more efficient (and probably more effective) community based approach.

Therefore, I would like to ask the Research 2.0 community to suggest some candidate platforms that either already support some of the above functionality or that could be used to build that functionality.  Notice the popularity of light-weight platforms such as doodle (when it comes to meeting planning) so perhaps we rather need a clever integration of light-weight web applications here as well (google docs + ???) rather than an  inflexible "dedicated" platform?

I welcome all your input (comments, suggestions, questions, ...), please do not hesitate to contact me.

Sincerely,
Pieter Van Gorp
r2ose2010 co-organizer
Eindhoven University of Technology


SHARE: slides for CASCON

Posted by: Pieter Van Gorp

Tagged in: Untagged 

Pieter Van Gorp

Hi all, Jean-Marie is so kind to present SHARE on my behalf. 

In the following slideset, you find some more background information and screenshots: http://www.slideboom.com/presentations/109003/SHARE:-Sharing-Hosted-Autonomous-Research-Environments-(IBM-CASCON)

Best regards,
Pieter Van Gorp


 

...

Concerning your presentation-slot, recall that 30 minutes are available for papers whereas 20 minutes are available for tool demonstrations.  From the organization of the GraBaTs workshops, I conclude that unless you plan properly, people will barely look at your actual software contribution during or after the workshop.  Moreover, in many cases the results from a particular plugin, algorithm or tool can hardly be reproduced some years after the workshop (think for example of the Eclipse version hell... or check http://csdl2.computer.org/comp/mags/cs/2009/01/mcs2009010005.pdf for an example from another community) Even in the context of the (small) Fujaba community, I observe that it takes people to much installation/configuration effort to quickly play with each-other's software.  


To counter this, we provide a means to install all software and data related to your paper remotely in a virtual machine (XP or Ubuntu based).  The supportive platform (SHARE) then enables others to reproduce your results in an optimally configured environment, without any installation/configuration hassle.  Additionally, if someone wants to extend your contribution (or report a bug), he/she can clone your image (if you grant access) and make the extensions back available to you.  Please go here to play around with the basic XP image and request a clone if you want "to SHARE" your contribution there.  It would enable others to experiment with your software before/during/after the Fujaba Days, which may make your presentation more effective...

That said, you can already download the Fujaba Days 2009 proceedings from http://is.ieis.tue.nl/staff/pvgorp/events/fujabadays2009/FDAYS09_proceedings.pdf
Please notify me as soon as possible when you find an obvious error: the proceedings will be printed in a couple of days.

Sincerely,
Pieter Van Gorp, Assistant Professor (UD)
    Information Systems Group, School of Industrial Engineering
    Eindhoven University of Technology (TU/e), The Netherlands
    Phone: +31 40 247 2062, Skype: pvgorp, Fax: +31 40 243 2612
    http://is.tm.tue.nl/staff/pvgorp/research/