Bug#825531: RFS: libcork/0.15.0+ds-1 [ITP]

[Gianfranco Costamagna <locutusofborg@debian.org>]

Hi,

>It's false positive.

ack

>> 2) lrwxrwxrwx root/root         0 2016-05-31 21:37 ./usr/lib/libcork.so.15 -> libcork.so.16.0.1
>The SONAME (15) and version (16.0.1) are given by upstream.
>I guess it's also for legacy reason, to be compatible with other system.
>So better not change by ourselves. I'll try to contact upstream, but
>it seems they're not very active, it may take time to let them change
>SONAME.


objdump -p debian/libcork15/usr/lib/libcork.so.1* |grep SONAME
SONAME               libcork.so.15
SONAME               libcork.so.15

it seems good, sorry I didn't check before.

actually libcork.so.16 seems soname 16, I didn't check carefully the above.


>Fixed now.


thanks!
>I'm not sure whether multiarch helps for this library.
>Since only shadowsocks-libev (in RFS now) will depend on it so far.
>I'll do it when the need comes.


well, it is just a matter of adding
"include(GNUInstallDirs)" to the cmake file, not sure why this isn't worth a fix

BTW this will allow 32 bit version to be installed on 64 bit systems, and will make porters
work easier.
(not asking you to patch, just saying that it can be trivially done, and upstreamed)


>I tested OK on amd64 and armel, by both common gbp and gbp with pbuilder.


yes indeed, I didn't see the tests between a verbose dh_auto_build and dh_auto_install.
I see them after checking again the build log


>Yes, kinda hacky.
>But original CDN url also may broken someday.
>User should read HTML without those JS anyway (I didn't test this).
>And the purpose to do this way is to avoid one lintian complaint.


not sure if lintian is right in this case, but feel free to leave as is


I can sponsor, just after some answers for the above.


thanks!

G.