Se ha denunciado esta presentación.
Se está descargando tu SlideShare. ×

Is there a future for Model Transformation Languages?

Anuncio
Anuncio
Anuncio
Anuncio
Anuncio
Anuncio
Anuncio
Anuncio
Anuncio
Anuncio
Anuncio
Anuncio

Eche un vistazo a continuación

1 de 16 Anuncio

Más Contenido Relacionado

Presentaciones para usted (20)

Similares a Is there a future for Model Transformation Languages? (20)

Anuncio

Más de Jordi Cabot (20)

Más reciente (20)

Anuncio

Is there a future for Model Transformation Languages?

  1. 1. Is there a future for MT languages? Jordi Cabot, Loli Burgueño, Sébastien Gérard @softmodeling – jordicabot.com July 2019
  2. 2. panel
  3. 3. To transform, or not to transform, this is the WRONG question - Shakespeare
  4. 4. Why are we having this discussion now?
  5. 5. An LSTM-Based Neural Network Architecture for Model Transformations – Burgueño, Cabot, Gérard, Models’19
  6. 6. https://mdse-book.com/ (Personal) perceptions on MTLS: • Research on MTL falling out of fashion • No new MT languages • Companies do not use them Let’s have this discussion as a community: True? Why? What next?
  7. 7. What have people said? (63 answers)
  8. 8. Average Expertise = 10.61 years
  9. 9. Simple Mature, standard
  10. 10. If you don't use any MT Language, can you give at least a couple of reasons why not? complex setup and IDE-specific dependencies transformation engines are mostly research prototypes No real debugging Hard to hire staff poor documentation ATL when simple transformations, Java when complex transformations complex to understand learning curve Practitioners without a background on MDE are skeptical performanceNot adapted to the web
  11. 11. Too much expectations BUT MTL concepts made their way into GPLs
  12. 12. Thoughts on the future of MTLs Lack support from easily comparable, strong, empirical metrics Scalability is a must An experienced tool developer may be quicker with an MT API than by learning a new language, and less dependent on proprietary tools Non-existence of a homogeneous community Little standardization of useful languages Undoubtedly needed and useful Need better showcase to industryMTs suffers from the misconception of modelling MTL are important. But we have to break out from classical transformation problems and approaches Not mature for industrial applications Stupendous and I hope they will be even more used MT may have longevity in Data Model Transformation We won’t be able to design a system without working with models due to its complexity MTs are overrated and over-researched
  13. 13. What do YOU say?: Is There a future for MTLs??? (and if so, what does this future look like?)
  14. 14. • Were you surprised by these results? • Do you think we have answers for the criticisms? • If they have been solved already, how to convince the industry? Is it a marketing or a technical problem? (e.g. MDE vs low-code) • Do we need a Yet Another MTL? Or we just stick to Java? • What should we do as a MT community (if anything)? • Do you plan to continue doing research on MTLs? • Should we just move to the next big thing?

Notas del editor

  • It’s an open discussion due to time constraints (goal: to maximize the participation).

    Feel free to interrupt at any time if you wish.
  • I think we all can agree that model transformation & manipulations is the core operation of model-driven engineering
  • On the one hand, we have been exploring whether AI will “eat” MT Languages.

    This started as a kind of a technical experiment but quickly triggered the discussion on how far we could go in this direction
  • On the other hand, I wanted to scratch my own itch and see if my personal perceptions were in fact a more common perception

    So, we (together with the ICMT chairs) decided to organize this open discussion
  • Good profile. Significant participation of industry. Plenty of answers and with a good expertise
  • On “Others” we mostly have custom & adhoc solutions

    Even at DAF, I wasn’t there but one of the people from the company confess they use Python!!!

    The 0% seems to show a disconnection between research and practice…
  • - This slide covers all the languages that people mentioned and some interesting comments about them
    People coincides that ATL is simple and QVTo mature and that is a standard
    For graph transformation, 1 person mentions tooling support while other says that there’s none
  • We see:

    Tooling maturity issues
    Scalability issues
    General MDE problems (convincing, finding experts,…)
  • TOO MUCH EXPECTATIONS. MT languages were rather 'new toys' 5 (or 10) years ago. New toys and technologies have the tendency to be tried out. Probably now the expectations (in MT languages) have not been satisfied..

    MT CONCEPTS NOW PART OF GENERAL LANGUAGES: Distinctive features of model transformation languages (e.g., traceability or automated rule application) will be taken up in general-purpose languages. + Modern languages increasingly provide functional programming constructs that mimic expressivity as found in OCL.

    So somehow MTLs have accomplished their mission
  • Most of the people support them! (although I’ve added only one box about this)
  • Are you surprised? Do you agree? Do we care? Do we just look the other way and continue as befoer?

×