Re: Heartbleed

["Randy Brukardt" <randy@rrsoftware.com>]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 3643 bytes --]

"Pascal J. Bourguignon" <pjb@informatimago.com> wrote in message 
news:87ha5q1b4l.fsf@kuiper.lan.informatimago.com...
> "Yannick Duchêne (Hibou57)" <yannick_duchene@yahoo.fr> writes:
>
>> That's because an Ada compiler is far more complex to design. This may
>> be another rational for a standard Ada minimal *frozen* *subset* (not
>> a new language) without every thing which is finally a derived form of
>> the most fundamental constructs with additional limitations.
>
>
> This rational could have been valid 20 years or 30 years ago, when
> proprietary software was more common, and free software was less
> numerous and of more difficult access.  Also, at a time, when a full Ada
> compiler would take 90% of the usual computer RAM, and 12 floppies.  At
> the same time, the first C compiler for Macintosh took only 2 floppies.
> And 30 years ago there were tens of different processors, and hundreds
> of different OSes, which made it very difficult to provide a compiler
> running everywhere.

That's certainly not true (of the full compiler size). The Janus/Ada 
compiler never took more than 4 floppies on any machine (it's in at most 
four parts, so it would have been impossible to run it otherwise), at least 
until the hard disk became the standard and the floppies were only a 
delivery mechanism. (And that was a lot more recent than you think. :-) It 
still only takes four PC floppies (Claw takes 4 more if distributed that 
way).

There were a few machines with microscopic sized floppies (I recall one that 
only held 90K), but I don't recall any compilers for any languages that 
worked on those floppies. Almost everybody required the 8" floppies (240K 
per disk in lowest density) or more capacity than that.

And the Macintosh was irrelevant at the time; almost all effort went to 
machines running CP/M-80 (including the Apple II emulator boards -- we sold 
several thousand compilers on that) and later MS-DOS. The full 
implementation of Ada for MS-DOS certainly didn't take more the 4 floppies - 
it could be run on a machine with only floppy drives (need two floppies, 
through, and the swapping made it impractical for long). If one had 8" 
double-sided floppies - 1.2M, that fit the entire compiler nicely. Indeed, I 
didn't get a machine with a hard disk until relatively late (when they got 
big enough to be practical); a pair of high-density floppies was more 
flexible in that one could replace one at any time and effectively have as 
much space as one needed. (10M hard disks had no such ability!).

> BUT  not anymore.
>
> Nowadays, you type apt-get install gcc or apt-get install gnat and the
> compiler is fetched from the Internet, and it doesn't make a difference
> if it takes 8 MB or 80 MB.

What's "apt-get"? It doesn't do anything on Windows or on my phone. ;-)

> Nowadays, there are only two processors: Intel on desktop and ARM on
> mobile, and there remains only ONE system: Unix (Linux on desktop Linux
> and Android, Darwin (BSD) on desktop MacOSX and iOS).
>
> And user machines have gigabytes of RAM and terabytes of hard disk, more
> than enough for any compiler and run-time.

True enough. It's no longer necessary to make a compiler that fits in very 
limited RAM. That being the primary expertise of RRS, we're no longer 
relevant. But so what? There are literally thousands of targets, once one 
considers all of the bare machine variants and the dozens of RTOS's in use.

You seem to think everyone can build on top of Linux, but certification 
requirements (and practical matters like what the customers are running) 
make that very untrue.

                                Randy.