Re: On the cost of libraries



Sander Vesik <Sander Vesik Sun COM> writes:

> I'm not entirely sure how to get a representative dataset for many
> of the present libararies.

It's very easy to do.  And the data doesn't have to be the same for
everybody.  However wants it can generate her own set and create an
optimized DSO and can send in the data to be pooled with similar data
from others to create the default reordering.

> Yes, this works very well for all kinds of plugins and similar, but not
> really for platform libraries. 

It works on any level you would want to have lazy loading.  An Xemacs
using gtk or terminal output can very well hide the function calls to
either of these backend implementations in some macros or so which
allow dlopen()ing.  It probably unrealistic to require gtk to make its
calls in the X libraries this way but nobody suggests that.  This is
no place where lazy loading would work.

-- 
---------------.                          ,-.   1325 Chesapeake Terrace
Ulrich Drepper  \    ,-------------------'   \  Sunnyvale, CA 94089 USA
Red Hat          `--' drepper at redhat.com   `------------------------




[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]