[phobos] Appender

Steve Schveighoffer schveiguy at yahoo.com
Tue Sep 7 05:09:07 PDT 2010


checked in changeset 1962 (opDispatch is cool!).

Note that this does not completely replicate the original functionality.  The 
original was unsafe for builtin appending, this one is safe, but may yield 
strange results.

For example:

auto str = "abcdefg".idup;
auto app = appender(&str);
str ~= "hijk";
assert(app.data == "abcdefg"); // on previous version, app.data would have 
"abcdefghijk"
app.put("lmnop");
assert(str == "abcdefglmnop"); // on previous version, str would read 
"abcdefghijklmnop"

Since we can't hook builtin array appending, RefAppender cannot tell when it has 
been used.  Any use of builtin append will reallocate anyways.

I'd recommend examining any code that uses appender(&arr) to make sure it 
doesn't also use builtin append on the array while using the appender.  Such 
behavior was unsafe for the original version of Appender anyways.

-Steve



----- Original Message ----
> From: Steve Schveighoffer <schveiguy at yahoo.com>
> To: Discuss the phobos library for D <phobos at puremagic.com>
> Sent: Tue, September 7, 2010 7:26:25 AM
> Subject: Re: [phobos] Appender
> 
> The old method is completely different from the new one, I'm not storing a 
> pointer to an array in the implementation struct.
> 
> I'll see if I can  wrap the appender functionality.  I'll add back the appender 
>
> signature  below, it will just return something different than Appender!(T[]), 

> which  shouldn't be a problem as long as you use  auto.
> 
> -Steve
> 
> 
> 
> ----- Original Message ----
> > From:  David Simcha <dsimcha at gmail.com>
> > To: Discuss  the phobos library for D <phobos at puremagic.com>
> > Sent:  Sun, September 5, 2010 10:36:26 PM
> > Subject: [phobos] Appender
> > 
> > Guys,
> > 
> > Can we add back the appender signature  appender(T)(T[]* arr) as a  
>deprecated 
>
> >function?  I'm trying  to compile some libraries (specifically  Orange) that 
> >depend on  this.  IMHO it was used enough by existing code that  it needs to 
>be 
>
> >deprecated rather than immediately removed w/o  warning.
> > 
> > --Dave
> > _______________________________________________
> >  phobos  mailing list
> > phobos at puremagic.com
> >  http://lists.puremagic.com/mailman/listinfo/phobos
> > 
> 
> 
>        
> _______________________________________________
> phobos  mailing list
> phobos at puremagic.com
> http://lists.puremagic.com/mailman/listinfo/phobos
> 


      


More information about the phobos mailing list