From: "Randy Brukardt" <randy@rrsoftware.com>
Subject: Re: Can Ada print coloured/styled text to the terminal? (ANSI escape sequences?)
Date: Wed, 8 Aug 2018 12:59:55 -0500
Date: 2018-08-08T12:59:55-05:00 [thread overview]
Message-ID: <pkfb2s$ulc$1@franka.jacob-sparre.dk> (raw)
In-Reply-To: LewaD.910619$aC7.586128@fx03.fr7
"Per Sandberg" <per.s.sandberg@bahnhof.se> wrote in message
news:LewaD.910619$aC7.586128@fx03.fr7...
...
> I think you put the finger on the problem with time, and that is that time
> implies change. And to require every newly written software to be
> compatible with every previous compiler does not work since that would
> mean freezing time. However to require new compilers to accept old code
> (sometimes with minor modification due to new keywords for example) is
> complete different story.
That only makes sense for code that was created to be portable.
That requires that the code was compiled on multiple Ada compilers, lest it
depend on incorrect implementations of Ada or vendor-specific
features/interpretations. By definition, that cannot be the case for Ada
2012 code, since there is no other such compiler. I was only worrying about
currently supported Ada compilers: ObjectAda (bit of Ada 2012, full Ada
2005), Apex (full Ada 2005), Irvine (full Ada 2005), and Janus/Ada (little
bit of everything, not a whole lot of anything :-)
The alternative is to say that every Ada compiler has to clone GNAT,
including all of its unique (and unimplementable in other technologies)
features like Unrestricted_Access -- rather than implement the Ada language
as defined. In that case, there is no real point to even having an Ada
standard or even other Ada compilers (they could never match up to the
"original").
That's not my idea of a healthy marketplace where the best
ideas/implementations have an opportunity to win. Nor is it practical for
Janus/Ada: for instance, changing the definition of Integer would break
binary compatibility for all of our customers (Sequential_IO, Direct_IO,
Stream_IO), which isn't a reasonable thing to do. Code sharing of generics
is not allowed in such a world (it's incompatible at the margins with GNAT),
meaning several man-years of re-engineering (and loss of capability, too).
And so on.
For a substantial portion of Ada users, GNAT lock-in isn't a problem. Then
use GNATCOLL and lots of other GNAT-specific software. For people that don't
want to be locked in, care is needed.
Randy.
next prev parent reply other threads:[~2018-08-08 17:59 UTC|newest]
Thread overview: 22+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-08-04 1:37 Can Ada print coloured/styled text to the terminal? (ANSI escape sequences?) Hazz
2018-08-04 1:59 ` Lucretia
2018-08-04 2:21 ` Hazz
2018-08-04 9:07 ` Jeffrey R. Carter
2018-08-06 6:35 ` briot.emmanuel
2018-08-06 22:15 ` Randy Brukardt
2018-08-07 5:12 ` Per Sandberg
2018-08-07 22:31 ` Randy Brukardt
2018-08-08 6:39 ` Per Sandberg
2018-08-08 17:59 ` Randy Brukardt [this message]
2018-08-08 18:28 ` Dmitry A. Kazakov
2018-08-08 19:23 ` Dan'l Miller
2018-08-07 8:33 ` Dmitry A. Kazakov
2018-08-07 13:59 ` Dan'l Miller
2018-08-07 22:36 ` Randy Brukardt
2018-08-09 17:08 ` G. B.
2018-08-07 0:01 ` Aurele
2018-08-12 20:56 ` Matt Borchers
2018-08-13 10:23 ` Aurele Vitali
2018-08-16 9:58 ` alby.gamper
2018-08-16 12:29 ` Aurele Vitali
2018-08-16 12:52 ` alby.gamper
replies disabled
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox