Personal thoughts about D2 release, D, the Universe and everything

Eldar Insafutdinov e.insafutdinov at gmail.com
Fri Nov 6 14:04:48 PST 2009


digited Wrote:

> I understand clearly that Walter follows his interests prior to any DIPs, complains, community-driven projects and epic NG topics with all sorts of arguments inside, so what follows is my personal opinion and random complains, maybe it will be of interest to anyone. Maybe even to Walter.
> 
> Starting D3 after D2 will be an epic fail, and no features will save D3 and D at all.
> 
> What fame and adoption could D have now, if after D1 release instead of moar features, W worked on rewriting D on D and IDE (small) on D, using a GUI toolkit like DWT?
> 
> D is for practice. So those, who practice really, write DIPs, complains, do community-driven projects and epic NG topics with all sorts of arguments inside, because when you start really using D, the sole compiler that passes some primitive tests is never enough. D even now is not suitable for real practice, even on windows. Incremental builds with template code, emitted randomly? Templates and OPTLINK? Many more?
> 
> If W after D1 release did D on D with IDE (small), using a GUI toolkit like DWT, he experienced it all and really could say that D is suitable for million-loc projects and is born from practice.
> 
> Just don't say that cross-compilation with GCC doesn't exist or D can't use C backends with C interface like GCC to port it to other platforms later.
> 
> One man can't do it. Even a colossus, even in 10 years. There are another colossuses in community, practicing D, that is mentioned to blow C++ away, with what DMD is written, so noone who likes D wants to dive into a C++ mess DMD source is. And if it was written in D?
> 
> If the language author separates himself from supporting the main community-driven projects (like std lib, package/install manager, GUIs and editors) and break those every crappy random "release", every newcomer sees the mess first of all, and the devs find supporting their projects to be a never-ending fixing. The language is its main parts, too, and the usage of the language starts from installing a compiler, finding an appropriate editor, setting up an std lib and trying to do some GUI. If every step of this chain is a pain in ass for even an experienced user, the features don't matter and the reaction to a programming language is bad. Creating a programming language for practice, the real one and against one that has exellent toolsets, is in providing a toolset for this practice, too (and first that a newcomer tries), so the author should not separate himself from this. Doing it all alone is impossible, but supporting and taking part is and should be done.
> 
> In D even today, *nobody* from language devs really cares about real practice and real adoption and *nobody* really practice it: even a small game and a specialized doctoral are very small and very different from the mainstream real usage.
> 
> 
> //----------------------------------
> 
> 
> D2 release is somewhere near, and the talk from W is "to improve the language (D3 with AST macross and other coolies) or to die". It's not alive now, really, it misses a stable and suitable toolset. D is now suitable for a small amount of tasks and requres a lot from one who wants to try it out. Start using it to see it yourself.
> 
> Dropping everything as it is now (crippled, unfinished or simply non-existing) and moving to another bright D3 future will square the entire mess (^2).

I could argue with some points of this post, but it is clear that D should slow down a little bit. I would like to see something like D 2.1, D 2.2, etc, so that they introduce language changes, but not in a drastic manner. Another moving target will negatively impact the user experience, especially it will confuse newcomers.

And yeah, the idea to rewrite dmd in D is just awesome! It will take some time, but it will ensure that dmd is a good stress test for itself.



More information about the Digitalmars-d mailing list