Re: Ada++

[raph.amiard@gmail.com]

Le vendredi 29 mai 2020 11:23:22 UTC+2, fabien....@gmail.com a écrit :
> On Friday, May 29, 2020 at 8:06:34 AM UTC+2, J-P. Rosen wrote:
> > Please provide a use case showing how these features are
> > necessary/important.
> 
> https://github.com/alire-project/alire/blob/dfa1e1e8029dee2959742b73ed8a0fc96e22c8de/src/alr/alr-commands-index.adb#L171

Or in GNAT: https://github.com/aosm/libstdcxx_SUPanWheat/blob/02812415a478d43bcc37a17bb779fcab146fbe4c/libstdcxx/gcc/ada/snames.adb#L59

Or in Libadalang:

https://github.com/AdaCore/libadalang/blob/1cf553d5fc37317c670888f0893bc25560c85b7b/ada/extensions/src/libadalang-env_hooks.adb#L47

Those are just two examples on the top of my mind. Every time you want to embed a multi line string in an Ada app you need to go through this frankly annoying gymnastics.

This is also annoying for compiler writers. In Libadalang we need to recognize those as special cases because they can create comb trees of unbounded depth.

> They are totally useless. [Embedding was never a good idea, embedded
SQL, embedded machine code etc.]

Demonstrably not. Anyway those kind of blanket statements tends to be false in general, I hope you can see that there are many legitimate use cases for this, and that the fact that you did not need it doesn't mean it's useless.