On Friday, 11 July 2014 at 15:42:04 UTC, simendsjo wrote:
On 07/11/2014 05:30 PM, Chris wrote:
(...)
Believe me, D's supposed sluggishness as regards GC is
not so important for most applications. I dare say 90% of all
applications are fine with the current GC.
(...)
I agree with this. The bottlenecks i my applications are MySQL
and
Microsoft Office (Excel, Powerpoint, or even just plain COM).
The same
bottlenecks as I get when using C#. Of course, it depends a lot
on what
you do, but for my use (and yours, and probably many others),
the GC
performance is something you can probably safely ignore.
Ah, but that's because you're comparing it to C#, not languages
that don't use GC. The big problem for D is that the market for
programming languages has bifurcated since D was created, with
the performant native-compiled languages like C/C++/Obj-C on one
side and the much larger market for easier to use but much less
performant, what used to be called "scripting," languages like
ruby/python/java on the other. Trying to be a better C++, by
borrowing some ease of use features like GC or reflection from
the scripting languages, leaves D stuck in the middle right now,
neither here nor there.
Who still uses native-compiled languages? Performance-sensitive
games, server applications that squeeze out performance, like
number-crunching or search engines, and desktop apps that need
the performance, that's about it. Everything else has either
gone to the web with a scripting language backend or mobile. I
hear that even enterprise LOB desktop apps are mostly written in
Java/C# these days, because they just don't need the speed of a
native language and can crank the code out quicker that way.
However, mobile could be D's saving grace, as native development
is back on iOS and even Android is moving to Ahead-Of-Time
compiling with the next release. Too bad D doesn't work on
mobile, even though some of us are working on getting it there.
D should focus on the native end of the market, by trying to be
the easier way to get most of the performance. You're not going
to get the scripting guys now, because native is just too hard
for them. If D can assert itself in that smaller niche of native
languages, it might have enough juice to go after the other end
later. I don't think either happens without a commercial
implementation, community development doesn't cut it. Linux
didn't take off till long after it got commercial vendors on
board, the same will be true here.
I don't mean to be pessimistic about D's goal of being usable by
all, from scripting to systems, as D may actually be good enough
to get there one day. I just think you're not going to get there
without focusing on taking over a niche at a time, particularly
the niche best suited to D right now, mobile.
