Re: How to get Ada to ?cross the chasm??

[Paul Rubin <no.email@nospam.invalid>]

Lucretia <laguest9000@googlemail.com> writes:
> This is truly not an argument either. Say someone's deciding whether
> to use C or Ada and there's this sub £/$/€ 1000 compiler for Ada. They
> will just go with C, because there's a fucking C compiler everywhere
> for nothing. Seriously, look at how many C compilers there actually
> are. There's loads of em, all free.

There's also non-free C compiler vendors like Rowley and Green Hills
doing good business.  I don't like anti-GPL whining as a matter of
principle (if someone can't handle the GPL because they want to write
non-free software, why do they expect a compiler vendor to give away the
store?) but I do sympathize with the issue that GNAT Pro is quite
expensive.  So I'm supportive of a $1K-ish product.  I can tell you even
as an unfunded developer, that $1K wouldn't be enough to sway a language
choice for a serious project ($25K is maybe another matter).

I was in some discussions with a hardware guy about an embedded project
a while back.  The compiler possibilities were:

   - a couple of commercial Forth compilers that were around $1500 for a
     one-time perpetual license with 1 year of maintenance, plus some
     annual fee if you wanted updates.  The hardware guy liked these
     compilers because they had good support for embedded development
     and he liked the language.

   - Any of a number of free Forth implementations.  This might have
     been workable.  It would have been used just for low level stuff by
     the hardware guy.  The higher level stuff would be written by me
     and I was interested in using Ada for that.
     
   - GNAT Pro (something like $25K).  We would have needed outside
     funding for this, or anyway some revenue from selling the product.
     $1500 would have been a heck of a lot more doable.

   - Gnat GPL -- I (as the software guy) was ok with this.  The goal was
     to sell a hardware product, and given its target market, releasing
     its internal software under GPL would actually have helped sell it.

   - GNAT FSF - didn't really think about this since I didn't feel like
     we needed to hold the code back.

   - GCC or some other C or C++ compiler - I was interested in Ada
     because of stronger safety properties, but in practice this might
     have worked out ok.  Rust wasn't really a consideration.

   - interpreted language du jour: didn't really consider at the time,
     but probably should have.  Even low powered embedded cpus can run
     them acceptably now.