Am 18.10.2012 um 13:09 schrieb Victor Rocco:

>> On 10/11/2012 6:13 AM, andy pugh wrote:
>>> However, as your component is a standalone module then the chances of 
>>> it breaking anything else seem rather small. The only danger seems to 
>>> be that it might break the build system.
>>> A resonable guide to the tests that you ought to run can be found here:
>>> http://wiki.linuxcnc.org/cgi-bin/wiki.pl?PreparingContributions
>>> You probably ought to check that the system builds cleanly with your 
>>> modifications in both simulator and normal mode, and that runtests 
>>> works.
> 
> I have problems with
> http://wiki.linuxcnc.org/cgi-bin/wiki.pl?PreparingContributions
> The last test "fakeroot debian/rules binary" fails. The problem is not with
> my code 
> because i have tested it with "master" alone, in normal and simulator mode,
> and it fails 
> with the same error.
> 
> Here is the procedure (for master and sim mode) http://pastebin.com/iSyWqhgQ
> And here is the error (last part) http://pastebin.com/nDazi5SS
> 
> I need some advice in order to know what to do.
> 
> Thanks you in advance,
> Victor.

I think there are really two issues:

(I assume this is on your local machine, not a buildbot log):

1) dpkg-shlibdeps: warning: symbol _Z10INIT_CANONv used by 
debian/linuxcnc-sim/usr/lib/libcanterp.so.0 found in none of the libraries.
... more of similar ones...

this looks like a c++-mangled symbol which decodes like so:
$ c++filt _Z10INIT_CANONv
INIT_CANON()

this symbol is actually defined, but not in a shared library dpkg-shlibdeps 
finds - and it's defined in two places for extra confusion:

$ nm ../bin/milltask |grep INIT_CAN
0805ac2c T _Z10INIT_CANONv

$ nm ../lib/python/gcode.so |grep INIT_CAN
00003982 T _Z10INIT_CANONv

canterp is a loadable interpreter, and it will be loaded if selected into 
milltask, or gcode.so.

Actually I think the dpkg-shlibdeps should find gcode.so and be happy; I dont 
know why it doesnt. 

Looking at debian/rules.in I see:
        dh_shlibdeps -l debian/@MAIN_PACKAGE_NAME@/usr/lib

probably adding the Python library path where it finds gcode.so will shut it 
up, I'm trying that now locally:

        dh_shlibdeps -l 
debian/@MAIN_PACKAGE_NAME@/usr/lib:debian/@MAIN_PACKAGE_NAME@/usr/lib/python


Short summary of long academic excursion: ignore the libcanterp.so related 
warnings for now, its not your fault - it's just that nobody went the extra 
mile so far to find the cause.


2) dpkg-shlibdeps: error: no dependency information found for 
/usr/local/lib/libmodbus.so.5 (used by debian/linuxcnc-sim/usr/bin/vfs11_vfd).

if you look at the Submakefile for vfs11_vfd, 
http://git.mah.priv.at/gitweb/emc2-dev.git/blob/refs/heads/master:/src/hal/user_comps/vfs11_vfd/Submakefile
you'll find that this is built only if LIBMODBUS3_USABLE is set; this is 
autodetected in configure, because we decided not to make libmodbus3 a package 
requirement because there's no 10.04 libmodbus3 package.

This happens because you have libmodbus3 installed on your system, but I think 
it should not appear once you push it through the buildbot.

to sumamrize: push it through the buildbot; you'll likely have the first 
category of errors again - ignore them, you're good to go - Seb has been 
putting up with them for years;)

The second error must have disappeared.

- Michael

> 
> 
> 
> ------------------------------------------------------------------------------
> Everyone hates slow websites. So do we.
> Make your web apps faster with AppDynamics
> Download AppDynamics Lite for free today:
> http://p.sf.net/sfu/appdyn_sfd2d_oct
> _______________________________________________
> Emc-developers mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/emc-developers


------------------------------------------------------------------------------
Everyone hates slow websites. So do we.
Make your web apps faster with AppDynamics
Download AppDynamics Lite for free today:
http://p.sf.net/sfu/appdyn_sfd2d_oct
_______________________________________________
Emc-developers mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/emc-developers

Reply via email to