Right now, what's the most important for the success and adoption of D?

[Regan Heath]

Bruno Medeiros wrote:
> I wanted to probe the D community of the following issue:
> 
> Right now, what would the most important thing for the success and 
> adoption of D? In other words, if one had to use D to develop a medium 
> or large scale software, what would be the most important things to 
> have, that we currently don't have (or aren't good enough)?
> (I'm not talking only about language features, in case you didn't get that)
> 
> I ask this on the onset of seeing several people commenting on how they 
> can't wait until we have AST macros, or how the feature X would be 
> awesome to have, while me, now that I actually started to some coding in 
> D (even if in a very small project), would actually rate those a much 
> lower priority versus other things I'd like to have in the D ecosphere. 
>  I'll do some more comments on this but first I'd like to hear about 
> other people's opinions.

IMO the biggest issue is libraries.

First and foremost is the phobos/tango thing (I realise the guys 
involved are working on this one, so this is not a dig!)

The next thing that bothers me is the many, many, many, ..., many, many 
different build systems and inter-dependencies that the available 
libraries for D use.

In my idea world I would be able to use DSSS (or similar) to obtain and 
build a library without having to first install:
   rake
   perl
   ...

and download;
   ..some 3rd party library..
   ..this other library..

and _then_ figure out how to correctly install, build and use those 
tools and libraries.

I once thought it would be cool to write a gui in D... after many 
frustrating hours download this, that, the other, and trying to decipher 
the arcane methods of building each of them I simply gave up.

Granted, I'm primarily a windows developer, not a hardened hacker* like 
many unix devs.  I'm not used to, nor do I like to, actually _build_ 
every tool and library that I want to use.  I would vastly prefer a 
simple binary download in almost every case.

*This definition of the word:
http://en.wikipedia.org/wiki/Hacker#Academic_hackers

Of course libraries which rely are essentially a C/C++ port will perhaps 
always be a bit complicated especially if you want the latest version of 
the C/C++ library - requiring you to download it and figure out how to 
build it etc - but even these could come in a semi pre-built form.

So, in many cases I reckon if we could all just agree to use a certain 
build process, with standard build tools (like dsss, bud, rebuild, etc) 
then life would be much easier for people like me who just want to get 
up and running in as little time as possible.

Regan