Re: Weakness of Ada is expensive / security / etc. ? Anything else?

[Jeremiah <jeremiah.breeden@gmail.com>]

On Sunday, July 26, 2015 at 3:41:19 PM UTC-4, Florian Weimer wrote: 
> > http://www.keil.com/
> 
> Here's a leaked price list:
> 
> <http://www.tagor.rs/files/cenovnici/Keil_pricelist_USD_2012.pdf>
> 
> Based on that, it appears that licenses cost several thousand U.S.
> dollars.  The product seems to come with support, though, but it's not
> clear for which time frame.  If you need the yearly fee option for
> continuous support, the pricing is roughly in AdaCore's ballpark.

The prices listed there are slightly higher than they are now (difference in prices is in the low hundreds).  The 1 year license is a "1 year of use" while the really expensive one is a "perpetual use".  If you want updates or support after the first year, you have to buy that separately even if you get the "perpetual" license.  The free version of this compiler only allows up to 2k of compiled code, which is pretty limiting.

I use a lot of Microchip parts...typically the PIC24 series.  C compilers range from the early thousands down to nearly free.  I use Custom Computer Services series compilers which run pretty cheap.  I by all of the different lines and the IDE at once for a one time fee of $600-700 and if I want the yearly updates, I have to pay $275 for the first license and $130ish for any additional licenses per year.  If you don't need all the different targets or the IDE, you can buy command line versions of the specific target lines you want (there are 4 different lines) for much less.  And you don't need the yearly updates if you have a good build (also, if there is a really bad problem, they will sometimes let you update for free).  The free compilers for these type of parts typically only have optimization limits but not code size limits like the Keil tools.

I long for the day I can get an Ada compiler for the 16-bit processors in the Microchip lines, but I doubt it will ever come.  They have a very different architecture from most GCC targets, so it wouldn't be as simple as porting an Arm or MIPS target.  The 32 bit parts are MIPS like, but they are not as well designed as the 16bit parts in terms of size and power consumption, which is a big deal in the part of the embedded industry that I work in.  I hate the feeling that it is just so out of reach.