Talk:Oberon-1

From Wikipedia, the free encyclopedia

You have new messages (last change).

Contents

[edit] Problems with the current aricle -- long list

I don't have the information to correct this, but

"It also offers an interesting feature which is only recently becoming available in more prominent development systems: exportability of procedures. Imagine that for some program you create a PNG viewer procedure and that you export it: any program on the system will be able to view PNG files merely by calling the viewer! That's true reusability."

is bogus. Code libraries predate Fortran, and shared libraries date back to or before MTS and the late 1960s. CORBA and things like it do this too. I'm sure there's something mildly unique, but I don't know exactly what it is. Likewise,

"The code will be both smaller than that in nearly all other programming languages..."

leaves me sceptical; it sounds like ad copy instead of an accurate statement. Oberon is in the same family as Modula-3, Ada, Pascal, and Java, and more distantly C, so it's not going to produce a huge difference in code size. Perl, Mathematica and APL will blow away any of the above languages in the right problem domain.

"...(mistakes will be less easily lost in a forest of detail)"

Which doesn't follow from the statement it's parenthetical to. Languages that add the detail of a variable definition for every variable save people from many mistakes that were made in Fortran. APL and Perl are both known for small code, and both are known for being hard to read.

"less opaque (fewer language 'features' mean less chance of confusing/misusing them)"

Love the scare quotes around features. How does removing enumerations and forcing people to use integer constants make things less opaque? In general, you have language features because otherwise you have to write a lot of code to emulate them, or because it's more efficent to have the compiler handle it. I had a professor who would honestly argue that programmers should just use a Fortran with extra features added until everyone was happy. Whether or not Wirth's minimalist strategy is good is still being debated, but the average programmer uses a programming language more complex than Oberon, and frequently much more complex than Oberon (C++, Perl, Common Lisp, Fortran 95).

"less subject to error problems from data typing problems"

C programmers might point out that strict data typing like Oberon has is a pain and can add to a lot of type changing, from integer to float, for example, and even worse if you want to start bitwiddling character input. An Ada programmer, like me, would point out that in exchange for that pain, you catch a lot of errors; but Oberon doesn't catch nearly as many errors as it can, and is less portable than it should be, because it doesn't include enumerations and ranges that catch errors and avoid depending on an implementation specific integer type.

This presents Oberon as the perfect language, and doesn't include any concept that it's not, that the design principles may have tradeoffs. This is not unique: Ada programming language doesn't include much discussion about the tradeoffs either, but mentions the features as facts, instead of saying how much better it is than "almost every other language". (comment by user:Prosfilaes, May 2004)

[edit] Sure it's enthusiastic. But worrying about POV is not needed.

The article is enthusiastic. There is evidence that simpler languages have greater reliability. The above comments are all true. Calling an enthusiastic article NPOV seems overreacting. The article could be more encyclopedic by incorporating the comparisons noted above into the article. Kd4ttc 19:57, 2 Jun 2004 (UTC)

If in the article on Christanity that "Christians were better people than non-Christians", that would slapped down instantly as NPOV. Writing that "Oberon programs are also less likely to be buggy" is no less NPOV, even if it brings up less emotional responses.
I've read about studies to prove that Ada is less buggy than C and C++, but I'm not going to add "Ada programs are also less likely to be buggy" to the Ada programming language page. If I was interested in doing so, I'd have to make it clear the details of the studys, which might give me the right to say that, in many cases, it appears that Ada can have a significantly lower defect rate than C and C++. If studies were made comparing Oberon and other languages, they need to be cited, and then you can say that in the instances of the studies Oberon performed better than the other languages studied.
I can't incorporate my corrections into the article, because I don't even know what it's talking about frequently. --Prosfilaes 08:31, 3 Jun 2004 (UTC)
Religion and politics get the strongest response so discussions on Mac vs Windows sure get exciting! (Politically incorrect comments on (delete groups) deleted) Wikins seem to be very tolerant. I think the comment on studies regarding correctness of programs would be very useful in Ada. The article on Eiffel could benefit there as well, with the interesting observation that commercial success seems inversely related to how clean the programs are, and that commercial success measures only sales of compilers, not utility of languages. The studies that discussed what goes into creating a reliable language seem to have been lost on a whole generation of programmers. Lucky grad students in the future, who can rediscover all those studies. Me: I'm just an old fart medical doctor. Not up to date on the studies we've discussed obliquely. I'd like to see the Algol families come back. I get ill just looking at C code. At least I'm in good company with Dijkstra feeling the same way. Once a port of something like Pascal, Oberon, or Modula port to Mac OS and is tightly integrated in the IDE I'll get entusiasm for programming again. Till then PHP, mySQL, and Apache will have to do. Kd4ttc 01:36, 4 Jun 2004 (UTC)

[edit] Libraries versus features

"It has likewise been argued, failure to include a feature may force the programmer to reimplement the feature in his code, causing many people to code the same algorithms, frequently with bugs or inefficently; however, this ignores the concept of libraries. Java is an example of a simple-enough language surrounded by an enormously rich class library."

First place, it's beside the point; Oberon does not have the rich class library. Second place, it's overstating your point; libraries have different strengths than features. There's no enumeration library that's going to do the compile-time checking and clarity and succinty of built-in enumerations. You can't provide templates in a library, either, and emulated templates aren't as type-safe as the real thing. Likewise, libraries don't give compilers as much information as features do; compilers will frequently give awful error messages with the STL because they're seeing the inside of the template rather than the feature.

Also, are you still a simple language if you have a Java-sized standard library? Implementations still have a long row to how, and it's not like it's any easier or less required to learn about a feature if it's stuck away in a library. --Prosfilaes 12:19, 27 Jun 2004 (UTC)

[edit] Smaller=Faster -- not necessarily

I'm posting to the talk page so maybe we can discuss a consensus, instead of back and forth changes.

"Oberon code is typically smaller, and so faster, than equivalent code in many other languages."

Is bullshit. It is absurd to say that Oberon code is typically smaller if you don't specify languages and projects. Likewise, smaller doesn't mean faster. In Code Complete, Steve McConnell shows several functions in C to calculate the integer log base 2 of an integer; the fastest function is the longest.

Or let's look at Hello, World.

10 print "Hello, world"

echo Hello, world

#include <stdio.h>
int main (void) {
  printf ("Hello, world\n");
}

MODULE HelloWorld;
IMPORT Out;
BEGIN
  Out.String("Hello World!");
  Out.Ln;
END HelloWorld;

Gee, Oberon produces the most code. Do you really think that means that the code will be slowest?

I suspect the thought here was object code (or machine code depending on the compiler) as opposed to source text. The Wirthian languages do tend to use actual words instead of one character symbols (or an implicit invocation) for such things as variable typing and such. Thus I suspect the strongly felt comment immediately above is skew to the intent of the article text. ww 20:27, 5 Oct 2004 (UTC)
How much object code is produced is completely compiler-dependent. Also, I don't read that and think "object code". I think source code. In any case, before it is put back in in any form, I'd like to see some evidence that it's correct. Even smaller object code doesn't always imply faster code.--Prosfilaes 04:33, 6 Oct 2004 (UTC)

[edit] Slim Binaries

Should slim binaries (roughly: dictionary encoded abstract syntax tree of code) get mentioned in the article? See [1] for a little further info on slim binaries. Zarutian 01:06, 22 june 2005 (UTC)

Bad link there. The argument seems to be something along these, not so explicit, lines. Snmall binaries have, all other things being equal, have fewer interdependencies and so lower complication and so fewer opportunities for bugs. Well, not intended ones anyway. Which is the general Wirthian point. True, it's possible to have slim binaries which have pathologically interconnected everything, but that's the situation several sigmas out, the unlikely case. After all, not even close to all C programs qualify for the Obfuscated contest.
It's a fair comment to make in WP, though hard to quite explain clearly. With all the refs and such WP tries to insist on. It's a craftsmanlike point, and equally hard to understand. In Pirsig's terms, it's a kind of 'quality'. That took a writer of Pirsig's quality to describe, I don't have such hopes of my own prose... ww 16:55, 14 July 2006 (UTC)

[edit] Proposal to Merge

I've proposed that the 'Oberon-2 article should be merged into this one. My reasons are --

  • Oberon-1 was never really much used, and needs only to be mentioned as a historical curiosity
  • Oberon-2 is usually called just "Oberon"
  • similarly, several articles that ought to link to Oberon-2 are actually linked to this one
  • most of the available documentation and implementations are for Oberon-2
  • the features that keep Oberon alive only really began with Oberon-2

If you look at the Oberon-2 article, you'll see it's quite short, because once you've discussed Oberon-1, there's not much more to say about it. So, without any negative feelings towards Oberon-1 or -2, I think these two articles should merge; and the Oberon-1 article is larger and better linked into Wikipedia.

What do you guys think?

OrangUtanUK 13:23, 10 August 2006 (UTC)

Makes sense to me. RainbowCrane 15:25, 10 August 2006 (UTC)

Good suggestion - assuming that it is set up so that anybody looking for information on 'Oberon-2' ends up directly in the right place in Wikipedia. Chris Burrows 23:58, 10 August 2006 (UTC)

[edit] Requested move

Oberon programming language → Oberon (programming language) – Conformance with WP naming conventions atanamir

See: Wikipedia talk:WikiProject Programming languages/Renaming poll

[edit] claim that Oberon is not object oriented

A recent edit by Taw included an edit summary noting the article said oberon was not object oriented. As this can be achieved in various ways, including that used by Wirth et al in the design of Oberon, I suggest that this edit is confused as to the underlying facts. And in any case, the article also says plainly the O is object oriented, just not in the C++, Java, or Smalltalk fashion. Comments? ww 05:22, 20 October 2006 (UTC)

I would say that while Oberon supports the practice of OOP, it also supports Imperative programming. It's a component-orientated language, which gives a lot more expressive freedom to the programmer. If I may go all poetical for a moment, the Copernican revolution in programming was the sudden awareness that algorithms should revolve around data and not data around algorithms - an allusion to the discovery that the earth revolves around the sun. But as modern thinkers we should be aware that in fact the sun and the earth revolve around their common centre of gravity, which turns out to be outside the surface of the sun. Likewise, you can't write a decent large software system without understanding BOTH the data and the algorithms. This is an archetypical Niklaus Wirth view - refer to his ancient book "Algorithms + Data Structures = Programs" (Wirth, 1975, ISBN 0-13-022418-9). Let us be glad that Oberon is not ONLY an OOP. OrangUtanUK 19:02, 20 October 2006 (UTC)
Oberon-1 wasn't even the slightiest bit object-oriented. In Oberon-2 there's no message passing, not everything is an object (in particular classes are not objects), polymorphism is limited to class hierarchies, and there's no object-based encapsulation. What it really supports is abstract data types and component-oriented programming (it is pretty much irrelevant whether it's "better" than oop, it's just not oop). It supports a little OOP, but to call it is too much to call it an "object-oriented programming language". Taw 17:51, 31 October 2006 (UTC)
I disagree - certainly as far as Oberon-2 is concerned. According to Hanspeter Mössenböck, the driving force behind Oberon-2, "Type-bound procedures together with type extension make Oberon-2 a true object-oriented language with dynamically-bound messages and strong type checking at compile time." Ref: http://www.statlab.uni-heidelberg.de/projects/oberon/kurs/www/Oberon2.OOP.html I recommend that you argue the case with him if you disagree: his contact details are http://www.ssw.uni-linz.ac.at/General/Staff/HM/. In the meantime, please rectify the changes that you made and discuss before making such changes in future. Chris Burrows 23:47, 1 November 2006 (UTC)

[edit] Merged Oberon-2 article in

I merged the Oberon-2 article in, following what looks like consensus and passive acceptance. Please feel free to vituperate me ... ! OrangUtanUK 18:49, 20 October 2006 (UTC)

I think they shouldn't have been merged. The article started with "Oberon is a reflective programming language" (reflection was added in Oberon-2), and then started to mix them randomly. Like here - "The design continued the Wirth tradition/strategy of attempting to simplify without loss of power. Oberon may be thought of as a Modula-2 with full object oriented class/object capabilities". OOP wasn't a design goal of course - it happened (to the extend it even happened, Oberon-2 was more component-oriented than object-oriented) many years later.

Maybe an article about both that cleanly separates them would be possible, but it's better to have two articles, than one that mixed things up so much. Taw 18:01, 31 October 2006 (UTC)

Taw: Please change the name of this page and all Oberon-1 references from Oberon-1 back to Oberon again (I would do it if I knew how). Oberon-1 did not, and does not exist. Chris Burrows 00:08, 1 November 2006 (UTC)
I concur with this request. There is an Oberon Report (for what is here referred to as Oberon-1), but not an Oberon-1 Report. The article is thus in error as to the nomenclature of its subject. An awkward position, that, for a WP article. ww 00:16, 1 November 2006 (UTC)
I have now logged a move request with the administrators for Oberon-1 to be chnaged to Oberon programming language. There are already redirects so I was unable to do it myself.Chris Burrows 12:32, 1 November 2006 (UTC)

[edit] things HAVE GOTTEN OUT OF HAND!

Noticing the problem (as above) as between Oberon-1 and Oberon-2 and all the merging and renaming and all, user:Mets501 attempted to help us with our confusion. There is now a disambiguation page for Oberon, pointing to the two articles. WP can endure a lot, but surplusage in disambiguation pages is too much.

We must delete the disambiguation page, rename the Oberon-1 article to Oberon and, if the merging is complete, delete the Oberon-2 article.

Merger folks, where do we stand on the merge...?

Ok, folks, let's get going... And, ...let's be careful out there. ww 03:26, 4 November 2006 (UTC)

The merging work was undone by User:taw. See the history pages. I'm not particularly concerned whether there is a separate page for Oberon and Oberon-2 (as for Modula and Modula-2), or whether the pages are merged together. But the 'Oberon-1' terminology HAS to go. Chris Burrows 22:51, 5 November 2006 (UTC)
I favour combining the two articles. It was me that merged them - the discussion pages seemed to agree that was the right way to go. User:taw didn't discuss his un-merging before he went ahead with it. I would like to make the following observations:
* Oberon(-1) and Oberon-2 are both known as Oberon. Oberon(-1) is no longer interesting, and needs only a historical footnote.
* The example code on the Oberon-1 page is actually Oberon-2.
* a reader wishing to understand Oberon-2 would need to refer back to the Oberon-1 page as well.
OrangUtanUK 14:53, 7 November 2006 (UTC)
Retrieved from "http://en.wikipedia.org../../../o/b/e/Talk%7EOberon-1_af5a.html"