Talk:Ada (programming language)/Archive 1

Ada Home

 * Ada Home - Old Web Site for Ada (not maintained)
 * I agree it is not maintained, but it is still used by Ada practitioners for reference. It is still recently referenced from the comp.lang.ada group. So I have restored the extlink. -Wikibob | Talk 11:24, 2004 Jun 16 (UTC)

Moved comments
'''Moved comments here from main text. Reorganized into threads.'''--buzco

Ariane 5
The descrition of the Ariane 5 Failure should not be on a side describing a programming language a link to a different side should be enough. Ariane 5 Failure

Hermann Wacker

I agree with that. it seems to me that this topic has nothing to do in an article describing a programming language all the more since it is rather a human error than a weakness of the language.

--Tarroux 07:28, 24 February 2007 (UTC)


 * It is relevant because Ada advocates claim that Ada doesn't allow these types of mistakes. That is, until someone brings up this incident or something like it, then Ada advocates say something else. Val42 23:21, 28 February 2007 (UTC)


 * No, we Ada-Advocates claim that Ada can protect against these type of error - but you also can switch the protection off if need be. The ability to switch the protection off is an important feature of Ada - otherwise Ada would not be usable for System programming. Krischik T 07:46, 12 March 2007 (UTC)

The statement that this incident caused the Ada runtime error checking to be questioned (I have been working with Ada for 20 years and have never heard this). This statement has no citation; the full report on the incident does not cite the Ada language as a contributing cause of the accident only the specification and design of the software. I believe this statement should be supported by a citation or removed. Richard A Brown 18:05, 9 March 2007 (UTC)


 * Yes, we do need a citation. This statement should be marked as needing a citation (if it hasn't already).  Then after at least two weeks, if there is no citation provided, it should be removed. Val42 06:01, 10 March 2007 (UTC)


 * We originally added this chapter because the Ariane 5 failure is often used by Anti-Ada-Advocates to spread FUD. Isn't it great that the Anti-Ada-Advocates can spread there false claims unhindered why we are asked to provide more and more citation. We have the full Ariane 501 Inquiry Board report on Wikisource, 3 References and 5 External Links on the main article - but no that is still not enough! Down with the truth, FUD rulez! --Krischik T 07:46, 12 March 2007 (UTC)

In any case, a human-error "let's just import the program directly" example isn't a criticism. "It's unweildy and unpopular" would be criticisms... and before anyone jumps down my throat, I don't know the first thing about Ada, so those were just vague examples. I'd love to see a criticism section like the one in C++ or COBOL, but until then, I'm just going to axe it and request an expansion. --The Centipede 20:00, 15 August 2007 (UTC)


 * I agree that this section needs to be rewritten. I have never understood the insistence on talking about Ariane 5 in this page.  True, this story has been used to criticize Ada, but I don't find it useful to litter this page with all sorts of badmouthing about Ada.


 * In fact (but this would probably belong to the flight 501 page) it would be more interesting to ask if there is a programming language design that would have helped in this instance. In C and C++ the effect of the conversion would have been "undefined" (meaning: anything goes).  In Java it would have saturated.  In C#, an exception would have been raised.  Now because this particular piece of software was not actually in use, not raising an exception would have been preferable.  But the important point is that there is no way that a programming language could guarantee some reasonable behavior after an unreasonable conversion.


 * There have been valid criticisms levied against Ada, and it would be interesting to list them and give a balanced view of their merits. But this is not what this section is doing. -- PascalLeroy 14:04, 20 August 2007 (UTC)


 * Tony Hoare: "For none of the evidence we have so far can inspire confidence that this language has avoided any of the problems that have afflicted other complex language projects of the past. [...] It is not too late! I believe that by careful pruning of the ADA language, it is still possible to select a very powerful subset that would be reliable and efficient in implementation and safe and economic in use." -- C. A. R. Hoare 97.87.57.112 (talk) 06:43, 29 November 2011 (UTC)

US Department of Defense
Really only used anymore in US Department of Defense stuff. Oh yeah, and Ada95, the most popular Ada compiler actually translates your code to C, then compiles the C. I love that.

--- Sorry, but that's a rather pointless and not only invalid but also uninformed statement, in fact you may be surprised that software compilation regularly involves translating high level code into ASSEMBLY language, which -surprise- usually happens to be even less-(strongly) typed than any conceivable high level language, so the point really isn't the target language/representation that a high level language is translated to, but rather how (strictly) this is done and how it is enforced that the corresponding rules are enforced. Thus, only because a compiler for a very strongly typed language (such as Ada) may -in fact, routinely- translate source code to a less-strongly or even virtually untyped target/intermediate representation (like C or directly assembly), this doesn't necessarily imply that the strongly-typed nature and goals of the original language is automatically affected and may thus be questioned: once compilation has taken place and succeeded, it doesn't matter for an Ada program whether its target language is C, Java, Pascal, Python or assembly, simply because until then numerous rules have been applied to the original code, ensuring to a very high degree that the underlying source code can in fact be properly transformed to (a subset of) the target representation. The real potential pitfall in such a scenario would in fact lie in 3rd party software that may translate the produced -valid- output to another target language possibly not applying similarly rigorous rules, however this issue is routinely addressed by settling for a well-defined and very limited subset of the target language which can be easily verified and validated using a corresponding test suite.

This approach is for example also taken by the SPARK language (a dialect/subset of Ada): Being basically a subset, SPARK programs are necessarily valid Ada, however they're being written with lots of meta information about expected and desired runtime behavior, this very meta information is provided in the form of comments in the source code, so it would not really affect an Ada compiler, a SPARK compiler however will happily process and honor such comments to parse, validate and augment the source code to come up with an accordingly enriched Ada program. So, this is again an example for a scenario where a comparatively strongly typed source language (SPARK) is translated to a less-strongly typed language (ADA) without necessarily compromising the safety margins provided by the source language and its specification.


 * Ada95 is not a compiler, it is a version of Ada, the most recent version, I believe.  I am also skeptical of the idea that Ada is "really only used anymore" in the US DOD.


 * Do you mean GNAT (GNU Ada Translater) when writing "Ada95"? This is no translater to C any longer but a frontend to the GNU Compiler Collection, as the C or the Java frontends are. By the way, it is _not_ used only by DOD staff and it is a highly capable language. In contrast to the popular misbelief it produces rather fast code (comparable to C).


 * Make that C, C++, Forth, FORTRAN, Java, Objective-C and Pascal front ends! May be others! --buzco


 * The team behind Gnat used in the beginning a proprietary Ada compiler and used that to build an compiler which implemented a subset of Ada big enough to support building itself. Now the team left that proprietary compiler in the dust. After that it was a simple matter of extending that they already have while using the base to build the next more evolved version.
 * Gnat until 3.11 or so (ca 1999) needed to generate a C stub for the sake of elaboration but that liability is no more !!
 * GNAT never was a Ada-to-C translator !
 * cfront was a c++-to-c translator !


 * SofCheck licenses its "AdaMagic" front end to a number of Ada vendors. There is one version of this front end that generates optimized ISO/ANSI C (though it is not widely used at this point).  There is another version that generates Java byte codes ("AppletMagic"(tm)).  The most widely used version generates a conventional compiler intermediate representation, and has well over 50,000 users.  The version that generates optimized C is useful because it allows the use of Ada on platforms that might not justify the expense of building a fully integrated Ada compiler.  The C-generating front end has enabled the use of Ada 95 on Crays, old Vax/VMS machines, old versions of the AIX operating system, 16-bit microcontrollers, etc.  We love that. ;-)
 * And actually, the USDoD has no particular connection to Ada anymore. The most loyal users of Ada are companies that like the low error rates that typically accompany developing in Ada, as well as those that are obligated to certify the safety of the software (commercial aviation, passenger rail, etc.).  209.6.248.170 14:20, 29 January 2007 (UTC)

See http://www1.acm.org/sigs/sigada/education/pages/myths.html for some answers to similar questions.--Parallelized (talk) 23:02, 1 February 2008 (UTC)

And: http://www.mil-embedded.com/articles/authors/gicca/ --Parallelized (talk) 23:12, 1 February 2008 (UTC)

theory books
Ada is used in a few compiler theory books because of its comprehensiveness and elegance.

Related websites:
 * http://www.cs.kuleuven.ac.be/~dirk/ada-why2.html
 * http://www.cs.kuleuven.ac.be/~dirk/ada-books.html —Preceding unsigned comment added by Parallelized (talk • contribs) 22:45, 1 February 2008 (UTC)

primary users
Should there be a section on primary users of a language like ada and how their requirements affected the language?

Ada is today still routinely being used for efforts where safety-critical software systems need to be developed, this is for example the case for aviation and space technology (think ATC, avionics, engine control etc) but also lots of other government software (think traffic surveillance, control and management), whenever lives or great financial resources are at stake, crucial software components are still being developed in Ada or even more rigorous subsets of it (i.e. SPARK). Likewise, for many such safety-critical software projects there are very high safety requirements to be met, which are often easier to satisfy by using Ada (critical software in railway control systems or even nuclear power plants, must have a well-defined and easily verifiable runtime behavior).

Also, Ada is also frequently the first choice when it comes to guaranteeing a high degree of maintainability for long-term software projects.

For additional in-depth resources, see the Ada wikibook and its discussion/talk page which features a "resource" section about Ada, its primary uses and benefits. —Preceding unsigned comment added by 82.83.66.83 (talk) 13:23, 21 January 2008 (UTC)

See this website for an overview of Ada users and application areas. Someone might want to incorporate some of this into the actual article. --Parallelized (talk) 22:42, 1 February 2008 (UTC)

Pascal
Ada has some tastes of Pascal still but not one of the bad ones. --Janet Davis

Well both Ada and Pascal are based syntactically on ALGOL 60, so that's not much of a surprise. Ada certainly took some ideas from Pascal, but fixed it's problems. —Preceding unsigned comment added by 68.222.42.172 (talk) 21:12, 18 January 2009 (UTC)

list of possible topics
(list of possible topics moved from main page) What about
 * how Ada tied into formal specification languages?
 * how Ada is very stongly typed
 * packages in Ada
 * concurrency support
 * real time/embedded system support ???
 * how Ada suffered from design by DOC committee/specifications and required super-fast hardware for compilation/debugging.
 * ?? others

list of possible topics
I think the Steelman language requirements page would be better as part of this entry, or maybe a subpage, as it isn't really of interest except in the context of Ada. Anyone agree/disagree? --Matthew Woodcraft

Development History: If possible a table summarizing the roughly 17-18 received, all but one Pascal-based designs, with names or references preceding the four finalist colors would be good. Any distinct features might be noted before their developers fade to dust. 143.232.210.150 (talk) 17:44, 14 May 2008 (UTC)

More Ariane 5
I don't think the note re the Ariane 5 disaster is quite correct. As I understand it, the fault was the re-use of a part and its software, which had worked properly on the Ariane 4. However, the more powerful engines on the Ariane 5 gave a thrust/velocity/displacement that was out of the part's design range, and the part detected that fact -- correctly, according to its original design. But since it thought it had a number that failed the sanity check it went into debug mode -- again according to design -- and started dumping debug data onto the rocket's control bus. So unless my understanding of the problem is wrong, the problem was a simple failure to review a part's specifications, and nothing at all to do with the programming language or compilation switches. -- B.Bryant


 * That's basically my understanding of the situation, except that 'detected that fact' is a bit strong -- it seems that the error was a CPU-level floating point trap, with no high-level-language handler. The Ada task in question didn't actually do anything useful after takeoff, so if the exception hadn't been explicitly suppressed the rocket would have been safe.


 * Certainly that bit of the article could do with improvement, but some mention is probably appropriate in this article (if it were taken out someone would just add it in again). -- Matthew Woodcraft


 * The Ariane failure occurred in a piece of the code that has no function in flight (only useful on ground). It was a simple run-time overflow of an float to integer conversion. Since the exception trapping and handling in Ada is somewhat heavy and it was thought that this overflow could not occur, no exception handling was present in this part of the code. In Ada, unhandled exceptions generated propagation to calling levels, this overflow condition propagated all the way up to the main line. Exceptions at the top level main line are handled with a CPU halt, with the assumption that a major hardware failure has occurred and a switchover to backup is needed.


 * The reason that Ada by some is claimed as a contributing factor for the accident is that such a failure occuring in low level was not contained there, but propagated all the way to the top level code causing a halt. Common system design practise in space systems required that malfunctions in subelements are not propagated to other parts of the system.


 * The article was misleading on this, I tried to make it more factual. Uffe 05:01, 2 September 2006 (UTC)


 * PS. The reference used is very bad third level hear-say analysis by a group trying to promote some software called 'Eiffel' and has no bearing on subject.


 * Note that it is only an extract from the main article. --Krischik T 07:51, 12 March 2007 (UTC)

Re: "design by DOC committee

 * Ada was not "designed by committee", each version (Ada and Ada-95) had ONE designer that had final say on any feature or change to the language. Its design is far cleaner and more organized than many other languages. -- RTC 01:38 2 Jul 2003 (UTC)


 * True. The DOD's choice of Ada was the result of a competitive tendering process in which Ada beat the other contenders.  The only role of a committee was to decide that it was better than the other proposed languages. -- Derek Ross

Re: "required super-fast hardware for compilation/debugging"

 * I added a section on the problems that the early compilers had and why. -- RTC 02:16 2 Jul 2003 (UTC)


 * Hmm, I never heard about Ada 80 and I discussed with lots of people involved in early Ada including Jean Ichbiah and people from the Ada/Ed project (first validated Ada compiler). Would you mind naming exactly the documents and compilers? Also point to document about elaboration issues? Thanks. Guerby 21:20 17 Jul 2003 (UTC)


 * I took the liberty of removing that entire section, not only because of Guerby's concerns, but because a quick search of the Web shows that the Ada Mandate was not signed into law until 1990, so it cannot be blamed for the quality of 1980-era compilers.
 * FWIW I have also heard mention that when Ada first came out it demanded things beyond then-current compiler technology, so maybe there is still a need for a section like the one I removed. However, if anyone decides to put such a section back in, please check and document any claims of fact included in the section. - B.Bryant 23:40 21 Jul 2003 (UTC)

Silly criticisms

 * Ada was widely called by many the "language of the future", but has never been used that much outside of the government. This title is often laughed at by programmers now. A major impediment to Ada's widespread adoption may have been that it is extremely verbose and the variable name conventions (requiring CamelCase and underscores to be used in the same variable name, such as This_Variable) are difficult to type, look ugly, and don't provide any information, unlike variable name conventions for just about every other language do.

I'm excising this whole paragraph. I've never heard it called the "language of the future", after reading a lot about Ada. For every language, you can find someone hyping it that looks silly in retrospect. Usage may be important, but it needs a lot more than just "never been used that much". "The variable name conventions ... are difficult to type, look ugly, and don't provide any information, unlike variable name conventions for just about every other language do" is a bit absurd; Ada has the same freeform variable names that most modern languages do, and the names I've seen in C code look much like the ones in Ada. To say that other language name conventions aren't difficult to type and don't look ugly is a very arguable case to make; BASIC and FORTRAN's short variable names, FORTRAN's type-name relations and Perl's type-name relations can be considered ugly and difficult.

I'd like to see some valid criticisms on Ada, but these are just silly. --Prosfilaes 02:19, 1 Jul 2004 (UTC)

date discrepancy?
In May of 1979, the Green proposal, designed by Jean Ichbiah at Cii Honeywell Bull, was chosen and given the name Ada

the above is chronologically the first mention of the name 'Ada' being associated with the Ada programming language. I'm quite certain that when I worked for the US Navy, and at sometime definitely prior to 1976 June 13, that I heard mention of 'the Ada Project' more than once. Do we have any info available from nonclassified source about this beingcalled the Ada Project, prior to 1979?

Reversion of Sep. 14th changes
There was nothing awful, but nothing great. Ada83 is one version of Ada, not another name for; "statically typed" is more unambigious than "strongly typed"; it was created to stop the proliferation of languages in the DoD, and improving on other languages (and not just C and C++) was a side-goal; what it was designed to do is important; and tasking is just one feature out of many.

I replaced "Ada was named after Ada, Lady Lovelace, the first computer programmer" with "Ada was named after Ada, Lady Lovelace", because the statement that she was a computer programmer directly contradicts the linked Wikipedia article (as well as other sources). My edit was reverted by UtherSRG with no explanation. I'm not trying to denigrate Ada Lovelace or the Ada language; I'm just trying to correct a factual error in a Wikipedia article. There is no historical evidence to support the common belief that she wrote programs for the Analytical Engine. -- BenRG 00:23, 12 Oct 2004 (UTC)


 * Sorry Ben. I get a bit overzealous guarding some articles from anonymous editors. You are right, the article on Ada Lovelace indeed points out that she probably didn't do any programming. Take a look again at the sentence in quesiton. - UtherSRG 14:00, 12 Oct 2004 (UTC)


 * Sure, I'm happy with that. -- BenRG 21:21, 18 Oct 2004 (UTC)

Ada Capitalization
I propose to change the capitalization of the identifier 'Text_Io' in the Hello World! example. In my opinion the correct letter case is 'Text_IO' and not 'Text_Io', as it's found in the Ada Reference Manual and also in the GNAT source code. Although Ada is not case sensitive this doesn't mean that capitalization isn't important. The Ada 95 Quality and Style Guide recommends to 'Use uppercase for abbreviations and acronyms', e.g. Text_IO (for Input/Output). Moreover, GNAT has an option to check attribute casing and reject any identifier that doesn't mach with its definition casing. Both forms are syntactically correct, but Text_IO is more correct and follows the language's guidelines of quality and style. Thanks -- suruena 14:13, 2004 Nov 30 (UTC)


 * Right!

--Krischik 14:51, 30 Nov 2004 (UTC)

JGNAT, MGNAT and A#
I've moved JGNAT back to the compilers subsection and added MGNAT (Ada .NET compiler). I've also removind the reference to A# to the related programming languages subsection (that was put there by myself). I'm not sure whether A# is only the name of the project, or there are any extension to the Ada language (other than 'limited with', a feature that will be part of Ada 2005). Are the languages accepted by JGNAT and MGNAT a subset of Ada? (Can someone confirm this?) In that case, maybe it would be necesary to add a section called Ada dialects.

NOTE: I do not consider SPARK as an Ada dialect because it greatly changes the language with data & information flow analysis (adding formal annotations), therefore becoming another language. But if A# makes any chages (I suppose those are minor changes), I will consider it a dialect.

-- suruena 23:38, 2004 Dec 6 (UTC)

More a superset
I would consider JGNAT and MGNAT more a superset. In order to support the virtual mashines additional featuers are added: Garbage Collector - a must, interface inheritance, limited with, Object.Operation.

Yes, some of which will be part of Ada 2005 - but not all with the same syntax. And on top ObjectAda did the Java extension differently. The hope is that with Ada 2005 all both will fall in line again.

As for A# - Tricky to say since MGNAT is the only A# implementation - so mayby there are just two words for the same thing.

SPARC and "Dialect" - I see the word itself broader then you - The are german dialects which I can't understand - but I can understand SPARC programms.

But if you are a native english speaker you may find a more appropiate word.

Anyway, I think the current wording is OK for far and I would leave it as it is.

--Krischik 08:43, 7 Dec 2004 (UTC)

Embedded systems
I removed the Embedded systems cat (a language alone isn't a system) and got a note regarding Ada's intended use within that context. While I wouldn't suggest re-adding the category, I would suggest that someone incorporate that fact in the article itself to better clarify that fact.--Hooperbloob 06:46, 14 Feb 2005 (UTC)

As a software engineer involved in the development of Ada, I would like to see included in here a comparison with C and C++. Ada is one of the few programming languages certified by the DoD because we produce much more reliable programs, more resistant to hacking than other high order languages, e.g., C++. However, where speed is concerned, you really want to use either Assembly Language or C, an intermediate language. Ada still requires a lot of memory and has a high overhead. There are some compilers that allow an optimized, bare-bones Ada but they need to be carefully checked. I suggest having two compilers and compiling the source code on both. You'll find different bugs each way.


 * I just wanted to start writing such a chapter but the first sentence in chapter "Features" allready reads: "Ada was originally targeted at embedded and real-time systems, and is still commonly used for those purposes.". --Krischik 18:19, 14 Feb 2005 (UTC)