[ksh93-integration-discuss] Re: [ast-users] /dev/tcp+/dev/udp notcompiled-in for Solaris ksh93r...
Dan Price
dp at eng.sun.com
Thu Sep 7 14:36:37 PDT 2006
On Thu 07 Sep 2006 at 11:28PM, Casper.Dik at Sun.COM wrote:
>
> >Some questions (CC:'ing Casper to have a look at the problem, too):
> >- AFAIK none of { Xorg, Xsun, Mozilla. FireFox, KDE, Kerberos5, bash }
> >uses libxnet... the question is: Why can't they use libxnet ?
>
> One of the biggest issue by folding the socket/nsl libraries, regardless
> of libxnet, is the huge pile of .data and .bss contained in libnsl:
>
> % size /lib/linsl.so
> 575446 + 33881 + 22251 = 631578
>
> Huge.
>
> >- The majority of applications in OS/Net uses libnsl+libsocket and not
> >libxnet (182 vs. 16) - is that just historical cruft that they do not
> >use libxnet (sometimes all three libraries are used (like
> >http://cvs.opensolaris.org/source/xref/on/usr/src/cmd/cmd-inet/usr.sbin/ping/Makefile#43))
> >?
>
>
> libxnet requires some different compilation; for the most part they
> are different wrappers around the same syscalls.
>
> When functions are made standard compliant they are sometimes generalized;
> that's why they're not fully compatible (different argument types:
> see sendmsg/recvmsg).
>
> Also you must note that libxnet is not a proper library; it's
> a filter; it does not contain actual code but rather a view
> on libsocket/libnsl/libc
Is there a even a policy statement to be made? Should new code use
libxnet? Should old code be cleaned up on an as-possible basis?
What changes would one have to make?
This is the first time I've been made aware of libxnet.
-dp
--
Daniel Price - Solaris Kernel Engineering - dp at eng.sun.com - blogs.sun.com/dp
More information about the ksh93-integration-discuss
mailing list