Re: Ada platforms and pricing, was: Re: a new language, designed for safety !

["Dan'l Miller" <optikos@verizon.net>]

On Tuesday, June 24, 2014 10:55:15 AM UTC-5, Shark8 wrote:
> On 24-Jun-14 05:25, Dan'l Miller wrote:
> Something where state-machines are first-class citizens seems like an 
> appropriate use for a DSL.

Ah, yes, domain-specific languages (DSLs) are yet another area that Ada202X could address as a claim to fame that would differentiate Ada202X from, say, C++ and Java and Swift, given that OCaml with p4, MetaOCaml, and Seed7 are already (imperfectly) blazing portions of that DSL trail.  Stage-n source code that teaches the Ada202X compiler a DSL at stage n to produce stage-n+1 source-code generation feeding into stage n+1 Ada202X compilation would put Ada back on the map.  Plus, full-fledged support for multistage programming via DSLs within an extensive multistage Ada202X compiler would likely solve not only the FSM problemspace well, but also could permit errors to be domain-specific errors to be detected at compile-time (at either stage-n or at stage-n+1 compilation) instead of at run-time, allowing Ada202X to one-up the dynamic-programming world.

For example, the dynamism of Objective-C & Swift at run-time would be moved to one of the later stages of multi-stage programming.

As another example, one of the A#1 uses of Objective-C's & Swift's *run-time* check of whether a method is implemented for a target object is for a GUI infrastructure and the app-domain usage there of to defer which object in a hierarchy of nested 2D widgets to process an event (e.g., mouse-click; button press; keystroke) at not necessarily the leaf widget and not necessarily the root widget/canvas, but rather any leaf-or-intermediate-or-root widget-node from leaf to root.  Some events are appropriately naturally discarded if no handler processes that event by the time that the leaf-to-roof walk completes for that containment chain of widgets at those coordinates (or in the case of keyboard:  the widget with focus).  But some events would be considered an error if no widget handled them during the leaf-to-root walk of widgets at those coordinates (or in the case of keyboard:  the widget with focus).  Hence, GUIs have a cross-cutting analogue of what is abstract versus concrete in OO polymorphism, but for GUIs the tree is not a tree of types of classes; rather for GUIs the tree is a containment tree as a likely-heterogenous n-ary tree of different types of widgets.  That eventual concreteness could be enforced across each leaf-to-root walk of a containment tree of widgets for a given window or for a given dialog box.  This is where a multistage compiler could really shine (and perhaps a multistage compiler with GUI-expressivity DSL capabilities), because it can analyze where any events that need to be handled are not handled along one or more leaf-to-root walks of that containment tree.  This is currently a run-time error-case in Objective-C and Swift, but need not be, because with the correct compile-time-philosophy-based programming-language such as Ada202X (and perhaps with a DSL among a multistage Ada202X compilation) has not been brought to bear on that problemspace.  Imagine putting Ada back on the map as the premier language for a whole new problemspace, mainstreaming compile-time checks for what has been until now various domain-specific run-time errors in all of the plethora of mainstream programming languages, which is precisely the special sauce that put Ada on the map back in the late 1970s and throughout the 1980s as the one to beat.