Andrei Alexandrescu wrote:
I figured I won't be able to unread the rest, so I stopped here
:o).
Well... my syntax must require a finer taste </posh> ;-p
Jonathan M Davis wrote:
On Friday, February 17, 2012 06:57:48 F i L wrote:
I knew someone was going to say that >:{
Well, you _were_ asking for it. :)
But I do honestly think that it's ugly. Obviously, different
people have different ideas of what a language's syntax should
look like - frequently strongly influenced by what language or
languages they used first. Personally, I _like_ the C style
syntax and do not understand why people want to get rid of
stuff like braces or semicolons (you didn't get rid of braces
in your suggested syntax, but it's a common suggestion for
people who don't like C style syntax). Overall, I think that
D's syntax is very good.
I started on C++ actually (ahh Codewarrior :)) and still find the
C (read D/C#) syntax to be the most readable overall, though,
after navigating the myriad of syntaxes and styles of web design,
I'm not too picky. But my syntax does have purpose behind it's
brilliance! For instance, all identifier names are placed at the
beginning of Lines, which eliminates eyeballing each line to find
it's definition and keeps accessor keywords (static, public, etc)
from getting in the way of variable grouping. 'case' is used
instead of 'if' to keep short, like-kind conditions aligned.
'else' can be used as 'else if' for the same reason. Contracts
are kept from bumping up against the definitions (important when
identifiers are first) and ending marks are largely unneeded in
the first place.
To eliminate Go's required "{" at the end of body definitions, I
would make each "free scope" require a scope keyword. This might
even be useful for something like functions derived of functions
(not thoroughly thought out):
foo( age:int )
{
before: scope {}
// foo code
after: scope
{
// default after code
}
}
bar( name:text, age:int ): foo( age )
{
// inherited foo code
scope( after )
{
// override after code
}
}
I know this is what polymorphism is for, but I think because
object methods are conceptually "random access" and scopes are
"in sequence" there might be a (any?) benefit to static function
inheritance. Maybe in combination with structs or UFCS? Idk,
maybe i'm just crazy :)
