svn commit: r227798 - in head: . lib/libpam lib/libpam/modules
kostikbel at gmail.com
Wed Nov 23 07:14:19 UTC 2011
On Tue, Nov 22, 2011 at 10:38:02PM +0100, Dag-Erling Sm??rgrav wrote:
> Dag-Erling Sm??rgrav <des at des.no> writes:
> > Jilles Tjoelker <jilles at stack.nl> writes:
> > > Although this will work, I think it trades the quality of the binaries
> > > for a cleaner build system. It is better to pass all libraries to ld(1)
> > > even though a .so may have unresolved references: the NEEDED entry
> > > serves as an additional protection against version mismatches and symbol
> > > versioning (if you ever add it) requires ld(1) to have access to the .so
> > > containing the definition so it knows the version name to store in the
> > > output file.
> > These are plugins. The names and prototypes of the functions they
> > export were set in stone 15 years ago.
> Sorry, that was a bit unclear. What I was trying to say is that symbol
> versioning is not and will never be an issue.
Does plugin depend on the library which loaded it ? In other words, does
it reference the symbols from the library ? If yes, then plugin _must_
have a dependency on the library recorded as DT_NEEDED.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Size: 196 bytes
Desc: not available
Url : http://lists.freebsd.org/pipermail/svn-src-head/attachments/20111123/b732bb5d/attachment.pgp
More information about the svn-src-head