Re: a new language, designed for safety !

[Simon Clubley <clubley@remove_me.eisner.decus.org-Earth.UFP>]

On 2014-06-10, Lucretia <laguest9000@googlemail.com> wrote:
> On Tuesday, 10 June 2014 13:28:33 UTC+1, Simon Clubley  wrote:
>> (The other reasons include the restricted set of host and target
>> platforms for Ada when compared to C and the fact the nicely
>
> the fact that the FSF version has been crippled on purpose to stop the
> building of certain said targets, you mean?
>

I'm aware of how much trouble you have had trying to get bare metal
ARM and bare metal PIC32 working because I looked at what you had
done a year or two ago. According to your comments and logs at the
time, it all seemed to be very fragile and prone to breaking in
various random ways with each new version of GNAT.

There's also the lack of public versions of Ada (or any Ada version
at all) for various non-mainstream platforms. For example, about a
year ago, I evaluated the state of gcc (including Ada) for VMS in the
public gcc kits. (ACT did a port of gcc to VMS for it's customers).

It didn't take me long to discover that not all the required bits
were there and that I would have a major task ahead of me. I managed
to build a C only cross compiler/binutils running on Linux which
produced VMS executables and was able to successfully compile binutils
binaries which ran native on VMS, but I didn't have any luck with Ada.

[Note: I didn't spend as nearly as much time on this as you have clearly
done as my interest in VMS outside of work has declined over the years
but I spent enough time to get a feeling for the size of the task. I did
post some detailed notes in comp.os.vms if you want to go looking for
them but they were more general gcc/binutils issues and not Ada specific.]

There's also the lack of Ada for microcontroller platforms other than
ARM (ACT supplied) and AVR (ACT/third party supplied). Just think of
all the stuff Farnell sells which doesn't have Ada compiler target
support, but does have C compiler target support. (I mention Farnell
because your website says you are in Leeds.)

>> pre-packaged ACT version of GNAT is licenced under the GPL instead
>> of the GMGPL.)
>
> This is a major problem actually. With GCC C/C++/.Obj-C/etc., someone can
> come along and just write a mobile app and sell it without any fear of being
> sued as you didn't release the source. Not so with ACT's GPL'd GNAT, all
> because they want people to shell out a stupid amount of money for support,
> which people writing mobile apps just cannot afford. The fact that they now
> release all their libs as GPL is just a joke as well; may as well say "we'll
> make sure nobody uses Ada for anything but trains, planes and bombs."
>

Back in the early days of the GNU project (~20 years ago) someone (maybe
RMS ?) suggested the output from the gcc compiler in general might be
subject to the GPL. It didn't take long for _that_ idea to be shot down
and to _never_ be suggested again. :-)

>> For general programming for which C would be used, if you want to get
>> your typical C progammer using something safer than C, I still think
>> my Oberon-14 idea represents the general path to take: create a
>> "safer", not "safe", language with some of the basic Ada concepts
>> transplanted into it and which is easy for a C programmer to learn.
>
> I have thought of something similar. I'll have to try to find your idea and
> take a look.
>

To stop you wasting your time, I only discussed it in comp.lang.ada in
the aftermath of the Heartbleed fiasco although I made some specific
suggestions in c.l.a.

Simon.

-- 
Simon Clubley, clubley@remove_me.eisner.decus.org-Earth.UFP
Microsoft: Bringing you 1980s technology to a 21st century world