I wonder how many people in this thread have VisualWorks and/or VisualAge on their machine as well as Pharo? They haven't pushed the user interface the way Squeak and Pharo have, and I personally find the VisualAge interface to be unpleasant. But they both offer a ton of support for connecting to other things. Let's take this list: - lots of good communication libraries into it yes and yes - spread it over Windows, Mac and Linux, yes and yes - make it open source VW and VA are proprietary, but you can get free versions (if you think I can afford to pay for stuff you are sadly mistaken) with all the usual access to the sources. - put some XML and JSON yes and yes - and solve printing, not quite clear, but I think the answer is yes and yes - multithreading/multiprocessing (framework) runtime AND (!) debugging, VW has recently made a lot of progress in that area - scripting it is not clear what you want here. - interconnections with other languages. yes and yes. C, and VA also talks to COBOL nicely. - Try adding a modelling and source code generator VW: very much so. VA: I have a lot of exploring to do yet, but I think so. - Build the whole stuff with concurrency in mind - offer specific data structure to help you here. Not clear what you want here. Smalltalk has had threads since at least 1980. If you built the whole stuff with concurrency in mind, you would not end up with Smalltalk, you'd get Erlang/Elixir. My Smalltalk has concurrent stacks, sets, bags, and dictionaries, and 'aCollection par collect:/select:/inject:into:/...' - Look for suitable persistency options. VW and VA both support serialisation. Pharo of course has Fuel.
The big thing that VW and VA offer that Pharo does not -- for obvious reasons -- is stability, with the quantity of documentation that stability makes possible. On Sat, 11 Jan 2020 at 05:15, Marten Feldtmann <itli...@schrievkrom.de> wrote: > > Am 10.01.20 um 15:42 schrieb horrido: > > > > >> So let's stop trying to convince people with things that mattered some > >> 20 years ago. Even the function point thingie we keep carrying in front > >> of our bellies (Capers-Jones was it?) is a lie when you want to build an > >> application for today's markets. > > > > I disagree that it's a lie. The study is based on thousands of projects and > > millions of lines of code over a period of several decades, including recent > > Well, naming it a "lie" is perhaps too strong - but Joachim (did you > have a bad Smalltalk day ?????) statement is from my point of view > correct - this talking about function point and productivity is an > academic point. > > If I am a Java developer and my productivity is around 10% compared to > Smalltalk developer it is still useful to use Java - because for my > problem there might exists already dozen of libraries and solutions. > > Using Smalltalk today is matter of personal taste and love - like many > other developers in other languages. > > Joachim mentioned the critical points and for me perhaps the following > statements are true: > > * Smalltalk development over the last decade ran in circles > > and due to that > > * Smalltalk is not solving the biggest problems any more > > So many time has been wasted to make a Smalltalk dialect running in a > browser. > > I would use (my loved) Smalltalk today only, if > > * I have an application, which was written in Smalltalk (and I have one) > > * Smalltalk is superior to other solutions in a specific topic (and with > Gemstone I have one topic) > > When I would start from scratch ... build a headless Smalltalk, put lots > of good communication libraries into it, spread it over Windows, Mac and > Linux, make it open source and put some XML and JSON and solve printing, > multithreading/multiprocessing (framework) runtime AND (!) debugging, > scripting, interconnections with other languages. Try adding a modelling > and source code generator. Build the whole stuff with concurrency in > mind - offer specific data structure to help you here. Look for suitable > persistency options. > > Go back to the time, where Smalltalk source code was hold in a > repository to manually work with it and and not getting software via > Github with some broken relationships between packages and nobody knows why. > > Use the browser (with Javascript) as the main UI and build a superior > interface in Javascript to the backend Smalltalk. Use the Electron > framework and build some specific support for Smalltalk into that. > > But even with that in mind you will not catch the Javascript developers > (because they are on that way already and they do not need Smalltalk), > but you may survive as a Smalltalk developer. > > Spread the word around, that multi-language development is a MUST and > one should support it. > > So, to summarize - this is my personal view of Smalltalk today - since > 1986, where I first met Georg Heeg on a Atari fair in Düsseldorf seeing > the first Smalltalk system in my life. > > Marten > > > > > -- > Marten Feldtmann >
