On 2017/07/30 14:52, Laurence Tratt wrote:
> With a snapshot from yesterday and fully updated packages (following
> Stuart's suggestion), libreoffice dies as follows:
> 
>   $ soffice
>   Warning: failed to launch javaldx - java may not function correctly
>   soffice.bin:/usr/local/lib/libicuuc.so.13.0: 
> /usr/local/lib/libicudata.so.13.0 : WARNING: symbol(icudt58_dat) size 
> mismatch, relink your program
>   soffice.bin:/usr/local/lib/libreoffice/program/libvclplug_gtklo.so: 
> undefined symbol '__cxa_thread_atexit'
>   ld.so: soffice.bin: lazy binding failed!
> 
> A quick grep suggests that maybe it's picking up the wrong C++ lib?
> 
>   $ grep -r __cxa_thread_atex /usr/local/lib
>   Binary file /usr/local/lib/libestdc++.so.17.1 matches
>   Binary file /usr/local/lib/libesupc++.a matches
>   Binary file /usr/local/lib/libreoffice/program/libvclplug_gtk3lo.so matches
>   Binary file /usr/local/lib/libreoffice/program/libvclplug_gtklo.so matches
>   Binary file /usr/local/lib/libestdc++.a matches
>   Binary file /usr/local/lib/libclangCodeGen.a matches
>   $ grep -r __cxa_thread_atex /usr/lib
>   $
> 
> I've tried rebuilding libreoffice from source (I forgot how long that
> takes...) but that hasn't changed anything.
> 
> Perhaps inevitably, abiword currently segfaults on load (but it's so long
> since I used it that I have no idea how long it might have been broken for).
> 
> 
> Laurie
> -- 
> Personal                                             http://tratt.net/laurie/
> Software Development Team                                http://soft-dev.org/
>    https://github.com/ltratt              http://twitter.com/laurencetratt
> 

Dirty hack and I don't know what it breaks, but if you need to run it
now, you can get LO to start up if you do this:

# chmod 000 /usr/local/lib/libreoffice/program/libvclplug_gtklo.so
# chmod 000 /usr/local/lib/libreoffice/program/libvclplug_gtk3lo.so

The icudata WARNING is not related (this isn't hit on a gcc build because
icu4c uses -finhibit-size-directive there, but this isn't available on
clang ... in any event *in this case* [but not in the general case] this
shouldn't be a problem.)


Reply via email to