[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
modules have sonames (again)
From: |
Felix Salfelder |
Subject: |
modules have sonames (again) |
Date: |
Tue, 22 Jan 2019 21:39:27 +0000 |
User-agent: |
Mutt/1.10.1 (2018-07-13) |
Dear all.
I recently ran into the situation described in a post from 2015 [1]. On
GNU/Debian, libtool sets the soname in a binary plugin (-module),
regardless of -avoid-version.
As a consequence, dpkg-shlibdeps issues warnings about unresolved
symbols. From dpkg-shlibdeps(1):
"""
the various warnings that you can encounter:
[..]
binary contains an unresolvable reference to symbol sym: it's probably a plugin
[..] The binary is most likely a plugin and the symbol is probably
provided by the program that loads this plugin. In theory a plugin
doesn't have any SONAME but this binary does have one and as such it
could not be clearly identified as such. [..] If the binary is really a
plugin, then disregard this warning.
"""
So I get the warnings. But they are slightly misleading and ignoring
warnings is not exactly best practise. Then, dpkg-shlibdeps keeps quiet
if there is no soname in the plugin, and no other change (tried with a
hacked libtool).
It seems apparent that a plugin does not need a soname. Non-libtool
build systems do not usually set it. So I wonder why libtool does it
that way, and whether I should wish for a change in libtool behaviour.
(I'd be happy to know about any other way to deal with the warnings. I
could not find it in the answers to [1], and not much has changed
since.)
regards
felix
PS: Please CC, I am not a subscriber.
[1] http://lists.gnu.org/archive/html/libtool/2015-11/msg00000.html
[Prev in Thread] |
Current Thread |
[Next in Thread] |
- modules have sonames (again),
Felix Salfelder <=