Description of problem:
expat 2.0.0 has been out for six months; it breaks SONAME relative to 1.9.x. An
upgrade to this might be worthwhile for FC7.
*** Bug 224178 has been marked as a duplicate of this bug. ***
The cost of this upgrade is high (20+ packages depend on libexpat.so.0 in Core
alone; a new compat package would certainly be needed) but the benefit fairly
low; I'm not sure it's actually worth doing this.
Do we have a problem that pyexpat is at 2.0 and apache is linked to 1.95.8 and
with mod_python we could have a clash? This seems to be documented here as a
, but I don't know for sure.
If pyexpat is embedding a statically linked copy of expat, that is a bug.
If pyexpat is embedding a statically linked copy of expat and exposing the
symbols to the global symbol namespace as-is, that is a really bad bug.
2.0.1 is out now. Probably worth an upgrade for F8, even though rebuilding 20
packages isn't fun.
The "20" number was pre-Extras-merge. Post-Merge, the number is 66 :(
Compat package prepared for review here in bug 251361, I'm working through mock
rebuilds of all dependant packages, to see if anything will break.
expat 2.0.1 is now in dist-f8. You are being CC'ed on this bug if your package
needs to be rebuilt against expat 2.x in Raw Hide. There is one minor API
change in this release, which maintainers may wish to check for:
- XML_GetCurrentLineNumber() and XML_GetCurrentColumnNumber()
now return unsigned integers.
The list of packages which still need to be rebuilt is here:
(based on the last Raw Hide push I got - so apologies for any stale data)
Have been rebuild (in dependency order where aplicable)
The following was rebuilt this morning:
qgis : http://koji.fedoraproject.org/koji/taskinfo?taskID=134051
k3d is being built right now, though it's taking forever.
xoo has been rebuild
avahi has been rebuild
gdb has been rebuilt (gdb-6.6-26.fc8).
libapreq2 being built: http://koji.fedoraproject.org/koji/taskinfo?taskID=135573
elinks has been rebuilt (elinks-0.11.3-5.fc8)
ser has been rebuilt.
Could you please comment on bug 131990 and let me know if it may be related to this one?
Please read bug 531990 not 131990 as stated in comment 18.