[mikewarot]

Wirth went too far with Modula, case sensitivity is a bug, not a feature. He became obsessed with purity, instead of usability.

Anders Hejlsberg was the author of the best Pascal implementation, greatly enhancing Pascal until he was seduced by Microsoft and helped start the evil that is .NET and C#.

Delphi was great until Borland decided to abandon most of their user base and pursue the corporate market.

Lazarus/Free pascal is really good, except for the abysmal documentation. The very lumpy approach to creating help actively prevents incremental improvements. There's no way to just fix one page of the help.

[Rochus]

> case sensitivity is a bug, not a feature.

Agree; that was the first thing I changed in https://oberon-lang.github.io/; besides the few academic oddities, original Oberon is a much better language than Pascal or Modula.

> He became obsessed with purity, instead of usability

There was definitely an academic bubble; for example, the claim that Oberon is a system language and can only be specified with 16 pages is demonstrably false, especially since there is a lot of code in the Oberon system that can only be implemented at all by means of (partially undocumented) backdoors to the language; unfortunately, these backdoors bypass the compiler's type checking.

> Lazarus/Free pascal is really good, except for the abysmal documentation

Unfortunately, the language is a huge, partly unspecified patchwork, where apparently all sorts of current fashionable constructs have been built in, partly even redundantly. The resulting complexity is hardly manageable with the present development approach.

[augustk]

Here are the mentioned sixteen pages if anyone is interested:

https://miasap.se/obnc/oberon-report.html

Oberon also has an interesting set of features to enable object-oriented programming without the syntactic sugar. Here is an article which describes basic abstractions in Oberon:

https://miasap.se/obnc/data-abstraction.html

[Rochus]

I referred to this one: https://people.inf.ethz.ch/wirth/Oberon/Oberon.Report.pdf.

[augustk]

The current version of Oberon (sometimes called Oberon-07) is about the same length though.

[Rochus]

But it's easier to count pages in a PDF than a HTML, and the 07 spec is even less complete than the earlier one.

[mikewarot]

I never thought of Pascal as a systems programming language. I always considered the inline assembler as the hack to get around that limitation. (Just as it is in C)

I am curious, what's the problem with redundancy? Having more than one way to do something doesn't seem particularly ominous to me. I'm sure you've got solid reasons.

[Rochus]

> I never thought of Pascal as a systems programming language

I referred to Oberon when stating that it isn't a system programming language, not Pascal; the latter is well suited for low-level system programming because you can take addresses and directly manipulate pointers, and there are facilities for multi-threading etc.

> what's the problem with redundancy?

FreePascal is at least three different languages in one; if you add compiler options the number of actual language versions even multiplies. There are e.g. different object systems which are even available at the same time in certain modes. This multiplies not only the learning effort, but also the development and documentation effort. It is apparent that FreePascal tries to follow C++ when looking at the feature set; C++ is a very complex language and so became FreePascal; the fact that not every feature could be freely designed but had to respect the existing feature set with all its interactions made FreePascal even more complex than C++ in some respects. And FreePascal has to carry on burdens from the past which were improved in Modula or Oberon, but which have to remain in the language for backward compatibility reasons.

[pjmlp]

Object Pascal was invented by Apple in collaboration with Wirth for Mac OS, as evolution from clascal used in Lisa.

Until the C and C++ took over inside Apple R&D, Mac OS was mostly programmed in Assembly and Object Pascal, and then it was time for C++ to take over.

Even the stuff that looks like C, is actually extern "C" under the hood.

Also no one really used the initial Pascal implementation, just like with C, most compilers had plenty of extensions, like Apple's Object Pascal.

Some of those extensions were later standardizes as Extended Pascal, however by then Modula-2 and Object Pascal (Apple and Borland's followup) were more relevant, Ada was coming into the picture, and thus everyone ignored ISO Extended Pascal.

[pjmlp]

> Anders Hejlsberg was the author of the best Pascal implementation, greatly enhancing Pascal until he was seduced by Microsoft and helped start the evil that is .NET and C#.

Some historical correction needed.

Many of the improvements in Turbo Pascal 4.0 onwards came from UCSD Pascal and Apple's Object Pascal, initially, and then followed up on interoperability with C++ for Borland products.

He wasn't seduced by Microsoft, rather by former Borland employees working at Microsoft wanting to refer him (those referral bonus), which he continuous refused until himself got pissed off with Borland's management.

"Anders Hejlsberg: A craftsman of computer language" interview

https://behindthetech.libsynpro.com/001-anders-hejlsberg-a-c...