DTIO, DTP, and Submodules

[Damian Rouson]

Hi Nick,

Hearing  your teaching experience is especially useful.  Thanks for sharing it.  

I wrote a 2D Navier-Stokes solver in C++ in grad school about 20 years ago and didn’t enjoy the experience — and that was before templates were widespread.  I’ve often wished someone would write “C++--" book that teaches only some subset of features that can be used safely without ever tempting the non-expert with the unsafe features, e.g., never using raw pointers.  In Trilinos, for example, the use of raw pointers is deprecated.  Instead, Trilinos developers are encouraged to use a reference-counted pointer class, which is of course a template class.  

If you know a good book on “safe C++,” please recommend it offline as I’ve diverged from the purpose of this mailing list.  I see O’Reilly has book named “Safe C++."


Damian


On Sep 25, 2014, at 9:55 AM, N.M. Maclaren <nmm1@cam.ac.uk> wrote:

> On Sep 25 2014, Damian Rouson wrote:
>> 
>> As I mentioned, I don't have a lot of use for generic programming in my work. I'm simply concerned about the future of a language that turns person after person after person away who makes the same suggestion. We can't assume that everyone else is either clueless or unaware of what they're doing. It's hard to see how a language community grows by shouting down a need that gets expressed repeatedly. It's really a shame that Paul was howled down on c.l.f. Fortunately for us, he's a committed contributor to Fortran through his work on gfortran. For every person with the commitment level of Paul, there are hundreds who will just walk away. I meet them in my teaching all the time.
> 
> I fully agree.  Unfortunately, bitter experience that shows that adding
> ill-fitting features just because they have important use cases ends up
> with a language which is, as a whole, close to unusable, even closer to
> unimplementable and completely impossible to teach, validate or often
> even debug.
> 
> I teach C++ as well as Fortran, and it has definitely achieved that
> state.  It CAN be used simply and largely cleanly, but you need even
> more self-discipline to do that than to program the comparable task in
> assembler.  Bjarne's Principles and Practice (which I use for teaching)
> is a worthy attempt, but you still need the self-discipline.
> 
> I believe that several recent extensions to Fortran have been steps too
> far in those directions, and the (rational) objections to templates is
> that it would be another, very big step of the same nature.
> 
> 
> Regards,
> Nick Maclaren.
> 
> 
>