home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.lang.ada,comp.lang.c,comp.lang.c++,comp.edu
- Path: east.alsys.com!news
- From: Ed Falis <falis@east.thomsoft.com>
- Subject: Re: C/C++ knocks the crap out of Ada
- Content-Type: text/plain; charset=us-ascii
- Message-ID: <31582A63.4BE9@east.thomsoft.com>
- Sender: news@thomsoft.com
- Content-Transfer-Encoding: 7bit
- Organization: Thomson Software Products
- References: <JSA.96Feb16135027@organon.com> <4ikbar$g0k@tpd.dsccc.com>
- <adaworksDoL573.7vs@netcom.com> <JSA.96Mar21160427@organon.com>
- <4iupk7$5t4@tpd.dsccc.com> <JSA.96Mar25205417@organon.com>
- Mime-Version: 1.0
- Date: Tue, 26 Mar 1996 17:33:23 GMT
- X-Mailer: Mozilla 2.0 (Win95; I)
-
- Jon S Anthony wrote:
- >
- > In article <4iupk7$5t4@tpd.dsccc.com> kcline@sun132.spd.dsccc.com (Kevin Cline) writes:
- >
- > > BTW, what architectures does ObjectAda support, and will ObjectAda
- > > code compile largely unmodified with Gnat, and vice versa?
- >
- > A Thompson guy could better answer this. I would be surprised
- > if Gnat and ObjectAda didn't compile things with little or no
- > modification. Heck, I've taken large chuncks of VAX Ada and moved
- > them to Gnat with no changes. Of course, these did not have any
- > OS specific stuff in them.
- >
-
- Well, I'll give it a shot. This year, it'll be Wintel, PowerPC/Win NT,
- Sun/Solaris, HP-UX and (I believe) PowerPC/AIX. We also expect Wintel
- cross to 32 bit X86 this year. Various PowerPC and 68K cross early next
- year.
-
- My experience so far is that the ability to cross compile code developed
- on GNAT or ObjectAda is pretty good - the main issues are in a couple of
- areas: one or other of the front-ends is a bit stronger in "corners" of
- the language, availability of identical bindings, and application use of
- implementation defined pragmas (Unchecked_Union definitely comes to
- mind, per discussion on another thread). On the other hand, there are
- an awful lot of issues that we used to see with Ada 83 compilers that
- just aren't there any more, especially such things as vendor defined
- unsigned types, low-level operations on addresses, shift operations etc.
- And these latter issues have always been the real ugly ones I've seen -
- where client code is calling
- "vendor_x_system.ought_to_have_been_defined_in_the_standard;"
-
- and it has to be reorganized or changed in hundreds of places.
- Obviously this was poor design, but Ada 95 will still help with its more
- uniform treatment of these kinds of issues.
-
- - Ed
- Ed Falis
- Thomson Software falis@thomsoft.com (617) 221-7341
- ========================================================
- Ideological disarmament: a koan for the 21st century
- ========================================================
-