Re: Why are Ada compilers difficult to write ?

["Dan'l Miller" <optikos@verizon.net>]

On Wednesday, June 20, 2018 at 2:37:31 PM UTC-5, Dan'l Miller wrote:
> On Wednesday, June 20, 2018 at 1:58:54 PM UTC-5, Luke A. Guest wrote:
> > Dan'l Miller <et> wrote:
> > 
> > >> * Tasks
> > > 
> > > So does C++17.
> > > 
> > >> * Protected types and objects
> > > 
> > > So does C++98.
> > >
> > 
> > Hardly. You can’t compare Ada’s high level tasking to the thin veneer over
> > threading that’s in C++. Come on.
> 
> Ummmm, apparently you are stuck in 1992 or something.  C++ ain't your father's Oldsmobile anymore, to paraphrase the old television commercials.  Time marched onward in C++, as shown in the timeline at the URL below.  Threads-of-execution, sequence points, thread-local storage, condition variables, multireader-uniwriter locks, threadsafe STL, atomic smart pointers, futures, latches & barriers, coroutines, transactional memory, task blocks (i.e., rendezvous), plus inheriting C99's and C11's work on multithreading and lack-of-aliasing.
> http://www.modernescpp.com/index.php/multithreading-in-c-17-and-c-20
> 
> Your point would be well taken if you were to shift to stability.  Ada pays especial attention to making sure that everything works together, whereas C++ plays the "undefined behavior" card too often, especially regarding multithreading interacting with other language features, most especially multithreading interacting with other undefined-behavior-swiss-cheese language features.  C++ having a language feature and C++ having a language feature that isn't like stepping gingerly through a land-mine field are 2 entirely different things.
> 
> > >> * Discriminants
> > >> * Elaboration
> > 
> > Oh I forgot that doctors and stir sections are equivalent to Ada’s
> > elaboration. Those sections just call the constructors and destructors.m,
> > it’s very primitive. You also can’t do complex initialisations in C++ like
> > in Ada.
> 
>   I did amend Jeffrey Carter's claim that Ada has a quite deluxe elaboration feature-set that very few if any other languages attempt to provide.  Perhaps you intended this reply to be inserted after some other topic instead.

I did *not* amend Jeffrey Carter's claim that Ada has a quite deluxe elaboration feature-set ...

> > > * Moderately generous reflection via 'Tags
> > 
> > Pick any C++ lib which hasn’t reimplemented this!
> 
> An arbitrary C++ lib in the C++ user community is not ISO14488 standard C++ in the compiler.  The OP is talking about compilers implementing language standards, not some arbitrary library out in the user community.
> 
> > > * Numerous aspects (historically standardized as pragmas)
> > > 
> > > So does C++11, C++14, and C++17.
> > 
> > Really? Which?
> 
> http://en.cppreference.com/w/cpp/language/attributes
> 
> 12 of them standardized so far up to C++17, more likely coming in C++20:
> alignas, noreturn, carries_dependency, deprecated, deprecated("reason"), fallthrough, nodiscard, maybe_unused, likely, unlikely, no_unique_address, optimize_for_synchronized
> 
> ... plus many more proprietary ones in GNU, IBM, and Microsoft C++ compilers.
> 
> Btw, with the exception of alignas and deprecated, C++'s attributes that have been standardized so far seem more hackerish than Ada's more pure-essence-of-reality aspects/pragmas.  But to the OP's question (and refuting Carter's claim), both C++ and Ada have aspects/attributes that need rather similar difficulty of implementation as popping up sporadically in the grammar, complicating what was a pristine much-simpler grammar decades ago.