Product Line Engineering

SPLC 2011 is getting close

Posted in Bits'n'Pieces by danilo on July 17, 2011

This year’s Software Product Line Conference will start in about 5 weeks from now (August 22nd) in Munich. While the program on the web page could be presented more user friendly, it contains a lot of interesting stuff. Many industry talks, i.e. valuable insights into real product line business; of course many tutorials. Two are given by  me: Requirements in PL and Migration to SPL. The tutorials are both on Friday, be aware that the numbering is counterintuitive:  T13 [Requirements] is in the morning andT12 [Migration] is in the afternoon.

Besides this, my company, pure-systems, will be present with several activities as part of its sponsorship, stay tuned for more information on this part of SPLC 2011.

PLE @ OOP2010

Posted in Bits'n'Pieces by danilo on November 29, 2009

Product Line Engineering is getting more and more mainstream: Next years OOP (25th – 29th January, Munich, Germany) has a 2 day track dedicated to this topic. Speakers include Jan Bosch, who I am quite sure, will give an entertaining talk on this views related to product line engineering. Other speakers include Markus Völter (Domain Specific Languages for PLE) and Peter Manhart (Variant Management in the Automotive Industry, in german) plus many more including myself (How to migrate to PLE successfully) 🙂 . So this might be the time to revive your knowledge of German (or English, see below) and have a few nice and interesting days in Munich.

  For those who have never heard about OOP before: This is the major German software engineering event for practitioners (1800+ attendees in 2009). Congress languages are German and English (majority of the presentations are held in German but quite a number are in English). It also includes an exhibition of tool vendors and consulting companies.

SPLC 2009: A Survey

Posted in Bits'n'Pieces by danilo on September 22, 2009

A few weeks ago  I was attending this year’s SPLC and, well, it was different this time compared to last year in Ireland. I think for the first time ever there were significantly less participants. But I had the impression that participants from the industry who were there, were more focused and seemed to be even more eager to get something out of the conference. So in the end, I think I liked it more than the one last year.

However, some of the trends in research seemed a bit odd to me. Take this proposition: “pure::variants is the most often mentioned commercial tool for product line engineering at SPLC 2009”. Based on this I did  a very simple survey on the usage of the term “pure::variants” and some other commercial tools in the proceedings of this years SPLC. Looking at the result I am quite happy with the survey: the term “pure::variants” is used 37 times in 6 articles. That sounds impressive. A little bit closer look reveals, of course, one of the articles I was the co-author. And the numbers do not tell you whether the respective paper was in favor or against use of pure::variants. So what is the value of this information? It is close to zero. You can get the  impression that pure::variants seems to have some influence. But that is about it. You have to read the paper to find out. And that is the point: A survey should give you some valuable additional information. Just numbers and charts are not enough. However, I must admit, I like the numbers about pure::variants very much, since it can be used for marketing, and who said marketing has to follow scientific rules 🙂

What else? Speaking about numbers and surveys: Kyo Kang, one of the master minds behind feature models, gave a keynote looking back at 20 years of history of feature models. And he also did kind of survey on citations of the paper that started it all (FODA – Feature Oriented Domain Analysis). But this was not part of the research track and the way he presented it, it gave some interesting insights beyond pure numbers. One them was that it took quite a number of years (almost 15!) until numbers of citations per year really rocketed (more than a hundred per year). That is aligned with my impression that when we at pure-systems started promoting feature models (2003) it was not so well known but soon after things changed. Again good for us 🙂

I liked the proceedings on a USB stick. First time saw this on a rather research oriented conference for me but I think quite a good idea. If only I could have gotten hold of one of those new eBook readers with electronic ink 😦 . 

Talking of equipment and first time experiences: I had 2 Windows blue screens on my laptop before my demonstration. Whenever I switched to external display, it crashed after few seconds. I have never seen this kind of thing happen before, and I have done a fair number of presentations in the last decade. Only the remote desktop stuff (and another laptop with a network cable) saved my day. The demonstration then went well but I do not have to have this ever again.

Lastly I have to talk about the location, which one of the things I heard often in discussions of the participants. First of all, I must admit I found the direct view on the runway of the San Francisco Airport from the hotel’s lobby bar quite entertaining. Watching planes landing in parallel, Boeing 747 slowing getting of the ground and this kind of thing were really relaxing things to do. Apart from that being in a hotel directly beside the airport for about a week is not as entertaining as one might think. And don’t get it wrong: The hotel as such was one of the best business hotels I have been to in the U.S.

SPL @ ECMDA 2009 After Show Report

Posted in Bits'n'Pieces by danilo on July 2, 2009

June (23rd – 26th) 2009 was the “Fifth European Conference on Model-Driven Architecture Foundations and Applications”  in Enschede, The Netherlands.

With very nice weather (27°Celsius compared to rain storms in not so far away southern Germany) and a very nice campus, it proved to be worth my and probably also the time of most the other participants. First for me was the MDPLE workshop on the 24th. Quite a number of participants (some 25) it seems to be a hot topic in the model driven community. Most of the presentations were on the technical side of the problem (to be expected), except for the key note of Mehmet Akşit, who made an very interesting proposal for checking a product line modeling approach: If you are able to capture the product line variability of the last 15 years in your approach, it is most likely also able to go with you for the next 15 years. The point was of course, that an approach must be able to deal with the future evolution. Since future is not known exactly (wouldn’t be a “real “ future otherwise), the only thing is to treat your history as simulation of the way the future will be. In many of the (earlier) approaches of model driven development and product lines as well, evolution and change was somehow left out of the equation. BTW, there were at least 3 conference papers looking at good ways to deal with model evolution (have a look at the Friday 10.30 session). As part of the workshop we had also a part where polemic questions were collected and answers for those discussed by the audience. This was also quite funny since there were some really polemic questions with a serious background.

Our own paper presentation  “Managing Flexibility: Modeling Binding-Times in Simulink” (Springer LNCS 5562) went well, but was slightly to long reducing the number of allowed question to just one 😦 . And in my demonstration of pure::variants for Enterprise Architect I tried to copy the wrong model (twice at least), contradicting somewhat the purpose of the copy action (which was to speed up the demonstration, 20 min are fairly short). Anyway, I was able to make my point and got some good response afterwards. Some other demonstrations also linked their model driven approaches/tools to product line ideas. IBM demonstrated an UML based approach, which was not to different from our integration with UML conceptually, so we cannot be completely wrong on this 🙂 .

The success of the MDPLE workshop makes me looking forward to the First International Workshop on Model-Driven Approaches in Software Product Line Engineering (MAPLE), which is obviously not the first workshop on this topic, only the first one using the acronym MAPLE :-). The workshop is part of this year’s SPLC in San Francisco (August 24th – 28th), on which I will post something later on.