On 30/05/2014 14:37, Dicebot wrote:
On Friday, 30 May 2014 at 13:27:10 UTC, Bruno Medeiros wrote:
On 29/05/2014 22:12, Dicebot wrote:
When similar question was asked during one of DConf talks vast majority
raised their hands as Linux users ;)
It is not that surprising considering D currently is most mature for all
kinds of services and server-side applications and this is domain of
Linux supreme uncontested rule.
That might have some influence (the kind of apps that people are
making), but I disagree that it's the main factor.
I suspect the poorer Windows D toolchain support is a bigger influence.
When volunteer effort is main development power actual use cases drive
toolchain enhancements, not other way around. It is not like someone
intentionally has made better tools for Linux just to make Windows
people sad.
Also native platform tools being open-source greatly helps in building D
ones on top. Remember that Walters article about adding 64-bit support
to DMD? He had to effectively reverse engineer object file format to
become compatible with Microsoft tools.
For sure, I wasn't saying the Linux D toolchain had more focus or more
effort put into. It's most likely the other way around even: the Windows
D toolchain had more focus and effort put into it!
It's just that despite that, it is poorer. And I agree, it's mainly
because of Windows being closed-source and proprietary that it is harder
to develop tool-chains. Especially for open-source developers (big
companies might have resources to be able to ask help from Microsoft
itself, but we can't).
I don't blame Walter for this, nor even do I think there's much more he
can do to address this in a reasonable time frame. It's just too much
effort for a single person.
I'm keeping my hopes on LLVM, and the work that is being done in LLDB
for Windows. In the meanwhile, using VisualD as a standalone debugger
might not be that bad, but it's still far from how better things are in
Linux.
--
Bruno Medeiros
https://twitter.com/brunodomedeiros
