Why is Ada considered "too specialized" for scientific use
in [ADA]
|
From: Denis McMahon on 7 Apr 2010 16:08 Charles H. Sampson wrote: > Has anyone written a paper "Ada for Scientific Programming"? I > envision such a paper as having all of the tasking-related stuff > stripped out and a heavy emphasis on the numerical issues. Probably the > distributed programming stuff could be eliminated too; I'm not sure. A > Paul Hilfinger comes to mind. A few years ago, as a scientist, I used and programmed for a complex physical simulation system written mostly in ada, but with some modules written in other languages (mainly old fortran or pascal models of physical systems that was interfaced to the simulation system). Ada is IMO quite suitable for scientific applications. Rgds Denis McMahon
From: BrianG on 7 Apr 2010 20:20 Georg Bauhaus wrote: > BrianG schrieb: >> Georg Bauhaus wrote: >>> Keith Thompson schrieb: (Sorry, I only meant to retain this paragraph:) >>>> Unfortunately, the C99 standard has not yet been universally adopted. >>>> Very few compilers fully support it. >> And this differs from Ada'05 how? > .... > > >> How many compilers support it? > > Fewer than the total number of compilers (Ada 95 or Ada 2005) > available, TTBOMK. > Is it more than 1? I don't remember hearing anything about other support. > >> More >> importantly (to me), how many non-compiler tools support it? > > Don't know. Syntax tools have few new things to deal with. > X-language tools might even be ahead if they had already supported > multiple inheritance of interfaces. Other tools for source code > analysis announce to support Ada 2005. Some makers depend on customer > demand and either fade or grow. > And some demand customers pay them to develop the upgrade, which the customer will have to pay for the privilege of using. If you want sustainable software, you can't rely on languages/versions that are not widely supported. It may not have been that many years since Ada'05 (what, basically 3 years, in essense?), but Ada'1z is already in work. When it's finished, what will the ration of '05 to '95-only tools be? Sorry, getting off of soapbox now. Maybe I ought to put wheels on it and take a ride downhill.
From: robin on 7 Apr 2010 22:01 "Simon Wright" <simon(a)pushface.org> wrote in message news:m2tyrnjc5f.fsf(a)pushface.org... | "robin" <robin51(a)dodo.com.au> writes: | | > As for your supercilious question, do I <<know of *any* algorithm that | > was first developed in machine code?>> -- | | Wasn't Ada Augusta's first program an algorithm to compute Fibonacci | numbers? That would certainly have been in machine code. | | And Alan Turing thought in machine code ... He did, because he wrote programs (including subroutines) back in 1945 when he designed the Automatic Computing Engine.
From: Robert A Duff on 7 Apr 2010 22:29 BrianG <briang000(a)gmail.com> writes: > Georg Bauhaus wrote: >> BrianG schrieb: >>> How many compilers support it? >> Fewer than the total number of compilers (Ada 95 or Ada 2005) >> available, TTBOMK. >> > Is it more than 1? I don't remember hearing anything about other support. Well, if you count each host/target pair as a compiler, there are quite a lot of Ada compilers supporting the full Ada 2005 language. All of those are based on GNAT, as far as I know. There are also non-GNAT ones that have partial support. - Bob
From: Georg Bauhaus on 8 Apr 2010 03:04
On 4/8/10 2:20 AM, BrianG wrote: > Georg Bauhaus wrote: >> BrianG schrieb: >>> Georg Bauhaus wrote: >>>> Keith Thompson schrieb: > > (Sorry, I only meant to retain this paragraph:) >>>>> Unfortunately, the C99 standard has not yet been universally adopted. >>>>> Very few compilers fully support it. >>> And this differs from Ada'05 how? >> > ... >> >> >>> How many compilers support it? >> >> Fewer than the total number of compilers (Ada 95 or Ada 2005) >> available, TTBOMK. >> > Is it more than 1? I don't remember hearing anything about other support. Can't speak for the makers, and my copies of non-GNAT compilers would need an update. However, some hints. Even a few years ago a non-GNAT front end had some messages saying something to the effect that "this feature is only available in Ada 2005"; I bet that the front end maker focused on analyzing program text more thoroughly than is needed in order to just compile it will have some support for the pre/post/inv features of 201Z, an important addition to the language IMHO. Likewise, it seems almost necessarily true to me that Janus/Ada has Ada.Containers. This is still speculation but will picture the Ada situation quite similar to what you find to be the case for other multi-vendor languages. |