<br><div class="gmail_quote">On Tue, Oct 13, 2009 at 9:44 PM, Don Stewart <span dir="ltr"><<a href="mailto:dons@galois.com">dons@galois.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<br>
<a href="http://www.galois.com/blog/2009/10/13/domain-specific-languages-for-domain-specific-problems/" target="_blank">http://www.galois.com/blog/2009/10/13/domain-specific-languages-for-domain-specific-problems/</a><br>
<br>
It advocates for Haskell + EDSLs, much as we have been discussing in<br>
this thread.<br></blockquote><div> </div><div> I am think that use of EDSLs for Physics (and similar science) are very arguable: <br> To use EDSL domain expert need to know language in which DSL embedded,<br>which is more difficult than learn just DSL.<br>
Not better, if EDSL use only subset of base language:<br> 1. because you need to teach this subset (probably rewrite of write new tutorials, books etc..)<br> 2. and if someone use few EDSL with different subsets of base language it can (and probably will) became mess.<br>
_So easiness in implementation results in burden for users_<br><br> I see such situation in Particle Physics where I am working. <br>All basic software: ROOT, Geant4 are actually EDSLs based on C++ <br>(and crippled C++: CINT). In my opinion this slowdown progress<br>
tremendously! I am know many physicists who don't know ever necessary basic<br>of C++, although they use ROOT and Geant4. <br> I am sure that what prevent them from learning C++ will prevent them from learning<br>any other general purpose language.<br>
_So my strong opinion that solution is only DSL not EDSL_<br><br> Roman.<br><br> </div></div><br>