Re: NSCharacterSet bloat?

[Richard Frith-Macdonald]

On 25 Jan 2006, at 19:11, Derek Zhou wrote:

> What I was puzzled about is why a smallish utility (handful of
> files), which propabably have little pratical usage other than in
> base, lives outside the base.

I think two reasons ...
1. the author of the utility originally did it that way.
2. the utility has no use at all inside base, so nobody would have  
changed it.

Let me elaborate on point 2 ... since this appears to be the cause of  
the confusion.
The characterset info translated by the utility tells you things like  
which characters are uppercase, lowercase, alphanumeric etc ... the  
unicode equivalent of the standard isupper(), islower(), isalpha()  
etc functions/macros.  As such it's not something that users of the  
base library ever need to change and there is no useful purpose  
served by having the utility in the base library ... it's of use  
solely when maintaining the base library in the event of a change in  
the unicode version that the library supports.

> Anyway, I can put all the relevent files into the NSCharacterSet dir
> under base and give you a tar ball and a patch to plug it in the
> overall makefile.

While I'd welcome improvements to the utility to automate checkiing  
of the unicode standard to see if it has been updated, I'd rather  
keep it where it is, since I see no point in adding it to the base  
library... it would just use a bit more disk space without adding any  
capabilities for users of the library.

> > I think it's better to stick to saying that you have a personal
> > aesthetic issue here ...
> Another benefit of seperation is to give users the option of updating
> charset data without recompiling base themselves. If gunstep goes
> mainstream, most likely people will get gnustep from distributions and
> refrain from manual installation (could mess up dependency).

But users shouldn't be updating the charsets (meaning that it's not a  
normal/reasonable thing to do) ... if they *really* want to change  
what characters are considered uppercase for instance, they can code  
in the changes they want ... use a category to override standard  
behavior, do all their checking via a function of their own etc etc.   
All the same sorts of options a user of the C libraries have if they  
want to change what ascii characters are considered uppercase with  
isupper()