I can understand the need for symbol literals in, say, LISP. There, a string was a sequence of bytes in a certain memory position. In this setting, string equality for instance, must be checked character after character. With symbols, one need only to compare the reference.
In the JVM however, strings behave more or less like symbol literals in this respect. Why do we need extra syntax for symbol literals in scala?
I don't see them used too much, in, say the Lift source, or various
The obvious advantage of removing a syntax, is reducing the complexity of the language, and maybe even speeding up parsing a bit (' can be decided to be a character literal without the need for lookahead).
Except of that, a big disadvantage for the current symbol literal syntax, is, that it confuses C-like syntax highlighters. So, say in Stackoverflow.com, scala code that contains symbol literal looks really bad.
(see also this discussion, I can understand why one would need a class Symbol that would be bound to contain only strings which are legal scala method names. However I can't understand the need for extra syntax. Symbol("method_name") seems fine to me for this use case).
I like symbol literals. I have no rational reason for it, but I like them.
On Wed, Aug 4, 2010 at 1:54 PM, Elazar Leibovich <[hidden email]> wrote:
I can understand the need for symbol literals in, say, LISP. There, a string was a sequence of bytes in a certain memory position. In this setting, string equality for instance, must be checked character after character. With symbols, one need only to compare the reference.
In the JVM however, strings behave more or less like symbol literals in this respect. Why do we need extra syntax for symbol literals in scala?
I don't see them used too much, in, say the Lift source, or various
The obvious advantage of removing a syntax, is reducing the complexity of the language, and maybe even speeding up parsing a bit (' can be decided to be a character literal without the need for lookahead).
Except of that, a big disadvantage for the current symbol literal syntax, is, that it confuses C-like syntax highlighters. So, say in Stackoverflow.com, scala code that contains symbol literal looks really bad.
(see also this discussion, I can understand why one would need a class Symbol that would be bound to contain only strings which are legal scala method names. However I can't understand the need for extra syntax. Symbol("method_name") seems fine to me for this use case).
On Wed, Aug 4, 2010 at 10:54 AM, Elazar Leibovich <[hidden email]> wrote:
> In the JVM however, strings behave more or less like symbol literals in this
> respect. Why do we need extra syntax for symbol literals in scala?
> I don't see them used too much, in, say the Lift source, or various
On Wed, Aug 4, 2010 at 11:57 PM, Raoul Duke <[hidden email]> wrote:
On Wed, Aug 4, 2010 at 10:54 AM, Elazar Leibovich <[hidden email]> wrote:
> In the JVM however, strings behave more or less like symbol literals in this
> respect. Why do we need extra syntax for symbol literals in scala?
> I don't see them used too much, in, say the Lift source, or various
On Thu, Aug 5, 2010 at 7:57 AM, Raoul Duke <[hidden email]> wrote:
On Wed, Aug 4, 2010 at 10:54 AM, Elazar Leibovich <[hidden email]> wrote:
> In the JVM however, strings behave more or less like symbol literals in this
> respect. Why do we need extra syntax for symbol literals in scala?
> I don't see them used too much, in, say the Lift source, or various
That's a good point for not using String in general for symbols. But I still don't see the need for a different syntax. We can use something like `case class Symbol(String sym) {sym.intern()}`, for symbols.
Locked
