Slashdot drew my attention to this ressearch ...
Slashdot drew my attention to this ressearch ...
http://ttendency.cs.ucl.ac.uk/projects/type_study/
An argument for languages like Delphi.
http://ttendency.cs.ucl.ac.uk/projects/type_study/
http://ttendency.cs.ucl.ac.uk/projects/type_study/
An argument for languages like Delphi.
http://ttendency.cs.ucl.ac.uk/projects/type_study/
Well, I didn't expect there to be a one-to-one correlation, but there still seems to be some relevance.
ReplyDeleteit's nice that these kinds of in-depth analyses have been done by computer scientists and are now fairly well known. The question is, why is Delphi so behind the times in terms of implementing things like this? Significant new features get added to languages that Delphi competes with in the marketplace while Delphi gets nothing but small tweaks. This "competition" isn't so much for the language, but as developers learn about these things and use them on other platforms, they come to Delphi and feel like they've just been transported back in time a decade or more. This doesn't inspire a lot of people to make an effort to encourage adoption of Delphi by the places they work.
These questions about the real benefits of strongly-typed languages arose back in the 80's when C++ was evolving, and the argument has always been that it's always better to let the compiler catch as many potential bugs as possible through such language features.
This article is nice because it begins to quantify the benefits in one specific case. They may not be directly transferrable to other languages like Delphi, but the principles still apply to some extent.
David Schwartz Several factors. One is that many people in Delphi don't wish for new features to be added to the language itself because they are fine with how it is.
ReplyDeleteWhich again has multiple reasons:
- one of them for sure is experiences how new features often were broken at first and only slowly got fixed and became stable (heck, generics are still broken in a very fundamental way when it comes to larger architectures where other languages just yawn at affecting compilation speed, IDE and tooling stability and binary size).
- those people that care about shiny and modern languages left Delphi or complimented their toolbox with those languages and enjoy them there
While customers not demanding these things is for sure an important factor I think the other is the lack of people with a vision and enough knowledge to actually design a programming language (no offense to any Embt dev). Right now I have to say the language itself is in a bad shape in terms of lacking official spec ("if it compiles its valid code" - except when there is a bug in the compiler and you have to argue with them if that is a bug, a missing feature or as designed - f*** that!). Also new "features" get slapped onto it turning it into some frankenstein monster. "We don't want to possibly break any code thus we don't add new keywords but put attributes on things" - but then there is no way to control the places an attribute is valid on. I could actually continue for hours about how much backwards compatibility crap the language is carrying around which makes absolutely no sense in 2017 anymore and is just some ancient relic of how things were when it was invented decades ago and all the inconsistencies in the syntax itself.
Probably also their compiler code base. I mean look at all the compiler bugs introduced and fixed in the past few years. And that is without adding fancy new features. Imagine the disaster that would possibly happen if they add big new things. Companies like Microsoft have much more R&D resources of course but only their new compiler architecture enabled them adding new features so rapidly and making them also immediately available to the entire IDE tooling.
Which is a terrible pity tbf because Delphi has some very unique features that some of the other languages don't even have.
Stefan Glienke Most customers who demanded new features already moved on to other languages.
ReplyDeleteI would guess that Delphi compiler is in such bad shape that it is almost impossible to add new features without enormous risk of breaking something. I remember some old Barry Kelly podcasts where he mentioned that the quality of compiler's code was very poor. And writing a new compiler from scratch is not an easy task for sure...