back

by Rochus·5y ago·view on hn ↗
The author is obviously enthusiastic. Smalltalk is undoubtedly impressive. However, many heroic sagas are rather marketing hypes than based on true facts.

> The Smalltalk galaxy was at the origin of the Graphical User Interface (GUI)

Well, the operating system of the Alto and the first groundbreaking WYSIWYG desktop applications were written in BCPL. Sure, Smalltalk did also run on the Alto, but the responsibles apparently have not considered it suitable for practical use. See e.g. http://xeroxalto.computerhistory.org/xerox_alto_file_system_....

> And since Java was strongly influenced by Objective-C, Smalltalk’s legacy amplified even more.

A bold claim. Actually Java was the implementation of the Simula 67 object model with a C syntax. Smalltalk was not involved here (neither Obj-C). Gosling is quite clear in this respect. The HotSpot JIT eventually was implemented by the folks who were already involved with the Self JIT, but it would be far-fetched to attribute these achievements to Smalltalk.

Even more points, but I'm running out of time.

5 comments
I am not sure if there were any other implementations of overlapping windows on the Alto other than Smalltalk-76 (which was what Steve Jobs saw in his famous visit to PARC). The windows idea was cleaned up and implemented in more static languages for the Star, of course.

As Dan Ingalls mentions in his recent HOPL (History Of Programming Languages) paper, several times an idea would be discussed between different groups and in the next meeting the Smalltalk guys would show it working while the other groups were still sketching on the whiteboard. Seeing the idea in action would prompt them to also implement it in Mesa or similar language.

Well, the claim was "Graphical User Interface (GUI)", not "overlapping windows". There are many examples of GUIs without overlapping windows and even usability experts claiming that it makes no sense to overlap windows. If I recall correctly neither Mesa nor Cedar used them.
In fact Smalltalk-76 didn't have overlapping Windows, and the people at Apple had already come up with that before seeing the Smalltalk demo. Bill Atkinson describes that here. https://www.cnet.com/news/tracing-the-origins-of-the-macinto...

The screenshot in the article (which _does_ have overlapping windows) is evidently Smalltalk-80 because Smalltalk-76 used the "hollow colon" character in its message syntax where Smalltalk-80 uses the typographic colon.

There might be a misconception. Have a look at Ingalls' 1978 paper "The Smalltalk-76 programming system design and implementation" (https://dl.acm.org/doi/10.1145/512760.512762) which clearly shows overlapping windows. There is even an online version based on the original source: https://www.lively-web.org/users/bert/Smalltalk-76.html
> Smalltalk is undoubtedly impressive.

Maybe ST80 was what Alan Kay needed.

Maybe what the rest-of-us needed was Modular Smalltalk?

"An Overview of Modular Smalltalk"

https://scholar.google.com/scholar?hl=en&as_sdt=0%2C5&q=An+O...

Actually Smalltalk-72 seems to better reflect Kay's vision of "object-oriented" than Smalltalk-80; see e.g. https://dl.acm.org/doi/abs/10.1145/3386335. Using tools like ENVY or Objectory you could do decent work in ST. Nevertheless, I switched to other languages like C++ and Java.
> Using tools like ENVY…

As many of us did.

What I had not previously considered was how the novelty of gui software and the vision may have overwhelmed more mundane concerns.

Seems like — even with the image, even with the ST80 language — there could have been a coarse separation into layers of functionality, and that would have been enough to address some of the practical concerns.

But that was mundane, not fun.

There were namespaces (aka environments).
My speculation was what it might have been like if PARC ST80 had a coarse separation into layers of functionality before release to the world.

If PARC ST80 had provided both:

no-gui no-ide boostrapped image + libs

:and:

bootstrapped image + libs + gui

:and:

bootstrapped image + libs + gui + tools

:as separate images.

Thanks, cool link!
"And since Java was strongly influenced by Objective-C, Smalltalk's legacy amplified even more."

If this was true, this would be a point against Smalltalk, not for it.

The influence of Smalltalk is generally overestimated, especially also in the referenced article. The author does not present evidence for his assertions, but simply the opinions of any people.
Yes. At the origin... as in XEROX Palo Alto Research Center, not the origin.
The origin of the graphical user interface was actually elswhere. The https://en.wikipedia.org/wiki/Semi-Automatic_Ground_Environm... system had a display system with a "graphical user interface" and a "ligth gun" (i.e. pointing device) by 1958; also e.g. https://en.wikipedia.org/wiki/Sketchpad had similar features in 1963; but these were CRT displays with vector drawings. The Alto in 1973 in contrast had a bitmaped screen and the first GUI following a desktop metapher (see e.g. https://www.microsoft.com/en-us/research/publication/alto-pe... or http://bwlampson.site/38-AltoSoftware/Abstract.html).