Re: KAI C++ Linux patch

[George R. <gr...@us...>]

Well, that isn't quite true.  We want ICU to become part of a standard 
distribution of Linux, but the vast majority of libraries on Linux use 
gcc.  We admit that there is still some more work in order to make it part 
of a Linux distribution.  When we build and test ICU we prefer a standard 
naming scheme that most people will prefer.  Some of these Makefiles are 
not even a part of ICU.

This problem with configuration choice is not unique to ICU.  Various 
Linux vendors configure and build their software as they see fit.  This is 
one of the reasons why there is now a Linux standardization committee so 
that the location and features of any given Linux distribution are all the 
same.

I'm open to suggestions to change the standard naming scheme of ICU.

George Rhoten
IBM Globalization Center of Competency/ICU  Cupertino, CA, USA

Sent by:        icu...@ww...
To:     George Rhoten/Cupertino/IBM@IBMUS
cc:     icu list <ic...@ww...> 
Subject:        Re: KAI C++ Linux patch



Ok, I see your point about what the intended usage for
ICU is. It seems that you envision that each
application that uses ICU will be very tightly bound
to it, and that the value of dynamic linking should
only be considered within the scope of the
application. That's reasonable and I'll happily use
ICU in that way and just statically link to it.
However, if that's all true, then there really is no
value in bundling ICU with a Linux distribution, since
it potentially could only serve such a tiny fragment
of users, namely those users that have been left
flapping in the wind by their application vendors with
regard to ICU configuration and compiler choice.