OpenEuphoria

Re: Just say 'YES' to strings

• Posted by "Kat" <gertie at visionsix.com> Jun 03, 2004
• 444 views

On 3 Jun 2004, at 4:07, Al Getz wrote:

> 
> Hello there,
> 
> Im not sure if you all are talking about strings in general or
> what, but what i'd like to see in Eu is this...
> 
> include cfunc_links.ew
> include string.e
> 
> atom x
> string s
> 
> s="Hello Mars"&0
> 
> x=c_func(xSetWindowTitle, {s})
> 
> Note in order for the above C function call to work properly
> the variable s would have to point to a memory location where
> the string is located.  The string gets there via the 's' variable
> assignment statement.
> 
> The thing about strings is if you know your variable is always
> a string then it not only makes calling C functions that require
> string arguments a lot easier it also means you can have better
> type checking.
> 
> For those people who dont like using this they dont have to
> of course, and can continue using allocate_string() instead.
> 
> It would be also possible to handle the string as both as string
> and a sequence.  All the usual operators would still apply, 
> except an Eu error would trigger if a program attemps to store
> a value greater then #FF or less then #00 to a string element.

That still leaves out multi-byte unicode letters.

> Examples:
> 
> string s
> 
> s="Hello there"&0
> ?s  --Hello there
> s=s[1..5]&0
> ?s  --Hello
> x=c_func(xMyCStringFunction,{s})
> 
> It may even be possible to set it up so that you dont have
> to keep using a terminating zero:
> 
> s="Hello"
> x=c_func(xMyfunc,{s})
> 
> although there would probably have to be a way to prevent the
> zero terminator in some cases.
> 
> Note in the above:
> x=c_func(xMyFunc,{s})
> 
> when the function name "xMyFunc" is linked the type 'string'
> is registered internally, so trying to pass anything other
> then a string would trigger a Eu type error.
> Example:
> xMyFunc=LinkFunc(MyDll,"MyFunc",{C_STRING},C_ULONG)
> 
> When a variable (such as s) is passed to this function during
> a function call, anything other then a string type passed
> as an argument triggers an error, just as any other type
> error would cause.
> 
> 
> Of course there are alternatives too.  A nice one would be
> to have an 'address of' operator that could be used to
> get the address of a string already in a sequence.

I used "address of" whatever in pascal a lot, ditto "pointer to" whatever, but 
RDS wants neither in Eu. It's handy for retypecasting, and indexing. Catch 
is, Eu does garbage collection, so updating the pointers after a collection 
could be time consuming.
 
Kat