Misplaced Pages

This is an old revision of this page, as edited by DNewhall (talk | contribs) at 23:00, 21 August 2006. The present address (URL) is a permanent link to this revision, which may differ significantly from the current revision.

July 2006: this page should describe the language!

A general look at this page (end of July 2006) suggests that it is mostly written from the viewpoint of comparison to other O-O languages and that a much better job can be done of capturing the real flavor of Eiffel. For example there is an inordinate amount of space devoted to trivial matters like typesetting conventions, but almost nothing about the core of the language: mechanisms such as multiple inheritance (mentioned, but no explanation of renaming etc.), once routines, constrained genericity, conversion, agents etc. Time permitting I hope to expand the article in the next few weeks so that in the end it will give a neutral but accurate picture of what Eiffel really is, from the viewpoint of people who actually use it on a daily basis. I also hope to add a few elements about the underlying methodology, without which it's hard to understand Eiffel, and to remove statements of opinion to replace them by more factual information.

Sounds great. I'll try to help when I can. --P3d0 12:23, 15 August 2006 (UTC)

As you can see I have started. It will take several weeks.Fuchsias 16:45, 15 August 2006 (UTC)

Older comments (2004 or so?)

This all reads a bit like advertising copy for the ISE commercial product. Is there someone out there who actually knows Eiffel, but also knows something else and can give a more realistic appraisal. (Yes, it does look like a nice language, but there are other nice languages, and this stuff is just over the top.) -- Geronimo Jones. (This comment is now largely obsolete due to much work on the page. Thanks guys.) -- Geronimo Jones.

I have taken out a lot of preachiness in the recent changes to the "fascism" section and tried to make it more factual; however, it was so absurd to begin with (and I am such an Eiffel advocate) that it may require more work.

Also, I removed unsubstantiated and inflamatory claims that Eiffel compilers are somehow "smarter" than C++ compilers. The discussion of moving runtime analysis to compile time presumably refers to the SmartEiffel compiler's ability to devirtualize method calls, and even to inline dynamically dispatched methods. However, this alone (while clever) is no great compiler breakthrough---Self and Smalltalk have had devirtualization for years---and it certainly doesn't allow SmartEiffel to claim that it's "smarter" than C++ compilers (some of which, in fact, can also do some devirtualization). On the contrary, SmartEiffel produces C code, and therefore leaves all the difficult aspects of compilation to the C++ compiler (eg. optimization, register assignment, etc.).

For what it's worth, I think the section Elegance, simplicity, or fascism? is really silly. There are a variety of languages that are simple and more high-level than C, and these comments could apply to any one of them. Moreover, "clever coding tricks" are not what I'd call optimization hints to the compiler; they are hand-optimizations. Finally, "Eiffel seeks to produce a quality software system over anything else" seems to imply that other languages are not designed for this purpose, which is silly — even C was not designed primarily for performance. Fascism is pretty over-the-top too — I dare anyone to design a "fascist" language. I'm seriously considering deleting this whole section, if there are no objections. Deco 21:13, 15 Nov 2004 (UTC)

I agree. The section is a sales pitch. Eiffel leans more toward clarity (over efficiency for example) than many other languages, but this section as it stands doesn't explain that very well. There are probably a couple of nuggets in that section that could be reworked into the article a different way, but on the whole I think the article would be improved by this section's deletion. --P3d0 14:38, Nov 16, 2004 (UTC)

Really, this article needs a serious cleanup. The trouble is, I have found that computer science articles don't stay cleaned up for long. Well-meaning CS people come along and add in their opinions, seemingly unaware that they are not objective facts. That's why I have turned my attention more towards science and astronomy articles lately.  :-) --P3d0 14:40, Nov 16, 2004 (UTC)

Pure OO

In the section where it is claimed that Eiffel is pure OO, it introduces control structures that are not in any way object oriented. I'm not a purist or formalist, but if you claim a language to be pure OO, "special form" control structures like loops and ifs should be replaced by message sends (like in Smalltalk). This page really is a sales ad!

Yeah, the claim of "pure OO" is meaningless, as everyone has different ideas of what constitutes purity. Probably the fact all Eiffel types are classes (even INTEGER and NONE) contributed to this claim, but as you say, there are different dimensions of purity. --P3d0 18:33, Dec 13, 2004 (UTC)

The loop example is silly, it would be better to show an example using the COLLECTION interface or at least *even* an example of that, since you almost always use that interface which is available in every kind of sequence/list/array/data structure.

from 
 x.start
until
 x.after
loop 
 x.forth
 x.
end

Have one question -- what is "x." you've used there? I'm not guru in Eiffel, but can you explain this for me? Why should this be placed in the article if this isn't clear? May it is better to use simple examples? -- Kolesnikov P.A. (ru-wiki) 17 June 2006

I guess that the x. is a typo, it has no mean in Eiffel syntax. Whereas x is a reference to an instance of a class that inherits from TRAVERSABLE (in the eiffel base). Momet 16:33, 17 June 2006 (UTC)

That's clear for me now that "x." has no meaning. But said "inherits form class in the Eiffel Base" -- you restricted that example only for those implementations with Eiffel Base. IMHO, that are ISE Eiffel and Visual Eiffel. Am I right in that? And SmartEiffel hasn't this class at all. --Kolesnikov Paul

I think I made a mistake, it's not TRAVERSABLE in the EiffleBase (from ISE), I guess it has to inherit from LINEAR, but don't trust me... There is something very similar in SmartEiffel, and it is ITERATOR, see this documentation for more informations Momet

Reflective?

The article began like this: "Eiffel is a reflective, object-oriented programming language" Eiffel is not reflective. I don't know where did that come from, but Eiffel is a statically typed, *compiled* language (it's usually compiled to C). The runtime system includes a garbage collector (usually mark-and-sweep) and not much else. Of course this doesn't mean one cannot write an implementation of Eiffel which provides facilities for reflection, but I haven't heard of such an attempt and it would be against the "nature" of Eiffel, as far as I can tell. Maybe we are talking about a different kind of reflection? Regarding programming languages it usually has the meaning defined at the famous ] but Eiffel is famous for inventing their own terms for everything (features instead of methods, etc).

Compiled languages can be reflective. There is no requirement that reflection has to be at runtime. Read Reflection (computer science) for more information. - DNewhall

Bias

Out of curiosity for the language I just read this entry, and I have got to say, though the author(s) clearly have been trying to be objective it is quite a biased account. The neutrality of the article feels very superficial, and the tone of the contents is clearly that of an ardent advocate. I had no preconceptions of the language but after reading the article I find myself slightly put off from Eiffel simply from the subtextual bias present the text. -- Mikademus

The group of people who have contributed to this article tend to be a self-selecting group of people who like the language, and may not even be aware of the biases we have inserted. We welcome any assistance in identifying the biased portions. --P3d0 17:56, 19 November 2005 (UTC)

It is difficult to point out specific things since the entire text emits a promotional air, not overtly, but undeniably there. It goes to the credit of the authors, however, that much effort has obviously been spent on trying to retain NPOV. Nonetheless, one salient thing is the "contra-sed contra" argumentation style prevalent in the text. To provide one example from early in the article:

"Eiffel intentionally limits stylistic expression, providing few means for clever coding tricks or coding techniques intended as optimization hints to the compiler. Some software developers feel constrained by Eiffel's simplicity and compiler-enforced structure; the language has been referred to as a "bondage and discipline" language.

In contrast, others feel that the simplicity of the language not only makes the code more readable, but also allows a programmer to concentrate on the important aspects of a program without getting bogged down in implementation details. Eiffel's simplicity is intended to promote simple, readable, usable, reusable, reliable and correct answers to computing problems. Eiffel seeks to produce a quality software system over anything else."

Though neutral by definition, the entire phrase "Some software developers feel constrained by Eiffel's simplicity and compiler-enforced structure" is designed as a counter-argument in itself, and the entire following paragraph is in turn a defence, then turns the table on the antagonist to show that the alledged weaknesses are really strengths, before finally, ending in close to pure promotion for the greatness of Eiffel. Again, I know nothing at all about the Eiffel language, but I do know marketing (even subconscious such).

This style is recurrant in the article, and, though from good intentions, is more reminescent of sales talk than encyclopaedic treatment or dispassionate analysis. Hope this helps!
Mikademus 21:04, 19 November 2005 (UTC)

I think the sentence fragment that reads "which emphasizes the production of robust software" is questionable. It would be better (more accurate/NPV) to change this to something along the lines of "emphasises the production of software which is provable correct to a formal specification". Esp. given the Eiffel tendency to fail hard on contract breaches at the expense of absolute robustness.

Thrown away Principles?

Does anyone know what exactly are these 'thrown away' principles?

This standard is not accepted by the SmartEiffel team, which has decided to create its own version of the language,
because they think the ECMA standard throws away important principles of the original language. Eiffel Software and
Gobo have committed to implementing the standard. Object Tools has not to date expressed a position.

--rolandog 02:08, 17 April 2006 (UTC)

Begun?

What's wrong with starting a sentence with "begun"? --P3d0 19:57, 9 May 2006 (UTC)

What code formatting convention should be used?

The font conventions and in particular the use of blue for program texts are part of Eiffel's style conventions. Whatever rules are applied in the descriptions of other languages are not applicable here. Thanks for respecting the specific rules for Eiffel. B-Meyer 17:18, 20 August 2006 (UTC)

That colour and style use would demand a short note before the first code example mentioning this. Also, in the discussion following or surrounding the text, the normal computer code tag should be used since using blue text in caps in body text makes the article look garsish, difficult to read, unencyclopedic and deviates from the wikipedia style guide. Also, it probably won't carry to a possible printed edition. So sure, go wild with idiomatic colours in code examples but use the conventional black-in-<code> style in the text. Mikademus 17:38, 20 August 2006 (UTC)

Thanks for the advice. I am applying it at the moment (include <code>), but I think that using black in the text would be confusing — the connection with the code extracts then disappears. Changing to code fonts, although this is not done in Eiffel documentation, should make the mix look less "garish" to you.

I think the blue text is improper for this article because A) it's easily confused for a link and B) no other article uses it. - DNewhall 21:10, 20 August 2006 (UTC)

These idiosyncratic blue fonts should really be removed. It doesn't make any difference what Eiffel's own documentation uses for formatting conventions: this is an article on Misplaced Pages, which has its own set of typographic (and other) style conventions. There would be a case for exact colors (well, as exact as web rendering can be) if this were on Color Forth or the like: i.e. something where colors are part of the actual program semantics. But that is not true of Eiffel. LotLE×talk 19:41, 21 August 2006 (UTC)

I appreciate the desire for consistency, but faithfulness to the subject matter should always supersede other criteria. Let me again ask you not to redefine Eiffel. The font conventions are part of the language rules. Eiffel is not just a programming language; it is a certain way of thinking about software and describing it. In particular, it is a language for publication of programs. B-Meyer 21:38, 21 August 2006 (UTC)

If the use of blue fonts is idiomatic to Eiffel code I for one have no issues with a blue colour being used in the code examples. Examples are just that: examples of language structure, grammar and style. Thus, though other may disagree with me, I personally find the blue colour appropriate if it is as is claimed, a convention used by the majority of the Eiffel community. However, I would for several reasons strongly recommend the blue to be removed from the bread text: first of all it does break with wikipedia style and will cause a lot of negative reaction from many readers and editors. Secondly it does in fact make the article a bit garish and distracting. Thirdly, colours in the text does not contribute anything, the point is already made in the code examples. Finally, as has been stated above, it is somewhat confusing because they blue passages as suggestive of hyperlinks. Look at this artificial paragraph I cut together from the article (note that I have link underlining turned off, as have many others too):

The concepts of Design by Contract are central to Eiffel. The keyword in that case is no longer require but require else.

All wikipedia programming articles follow the convention of black-in-<code> for keywords etc discussed in the text. That this article should not will ruffle feathers and is, above all, unnessesary and superfluous. When I look at a painting I want to see the colours. When I read a critical review of it I do not want it written in matching colours. Also, think about the wether to use bold and italics in the text, the current libera use, while mirroring the constructs in the code examples, are again if not confusing then at elast contributing to the garishness. Just plain, unadorned code will do in the text regardless of example formatting. Mikademus 22:17, 21 August 2006 (UTC)

Adding another note when on the topic, every so often there are links inside code-tags, like #include <iostream>. Considering that, I assume you see the unreasonableness of colouring the text itself. Mikademus 22:35, 21 August 2006 (UTC)

Two things, first the statement "Let me again ask you not to redefine Eiffel. The font conventions are part of the language rules." by Dr. Meyer is incorrect. Quoting the ECMA standard : "The color-related parts of these conventions do not affect the language definition, which remains unambiguous under black-and-white printing (thanks to the letter-case and font parts of the conventions). Color printing is recommended for readability." So color printing is NOT part of the standard for any other reason besides formatting of text. If this is to be considered part of the standard then we also have to use BNF to describe everything because that is defined in the standard. Second, while Eiffel is defined as a specification language too a quick look at the title of this article shows the text "Eiffel programming language". - DNewhall 23:00, 21 August 2006 (UTC)