summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
-rwxr-xr-xindra/newview/viewer_manifest.py19
1 files changed, 18 insertions, 1 deletions
diff --git a/indra/newview/viewer_manifest.py b/indra/newview/viewer_manifest.py
index d65d51ce32..8da1be2769 100755
--- a/indra/newview/viewer_manifest.py
+++ b/indra/newview/viewer_manifest.py
@@ -1044,7 +1044,24 @@ class Linux_i686Manifest(LinuxManifest):
self.path("libalut.so")
self.path("libopenal.so", "libopenal.so.1")
self.path("libopenal.so", "libvivoxoal.so.1") # vivox's sdk expects this soname
- self.path("libfontconfig.so.*")
+ # KLUDGE: As of 2012-04-11, the 'fontconfig' package installs
+ # libfontconfig.so.1.4.4, along with symlinks libfontconfig.so.1
+ # and libfontconfig.so. Before we added support for library-file
+ # wildcards, though, this self.path() call specifically named
+ # libfontconfig.so.1.4.4 WITHOUT also copying the symlinks. When I
+ # (nat) changed the call to self.path("libfontconfig.so.*"), we
+ # ended up with the libfontconfig.so.1 symlink in the target
+ # directory as well. But guess what! At least on Ubuntu 10.04,
+ # certain viewer fonts look terrible with libfontconfig.so.1
+ # present in the target directory. Removing that symlink suffices
+ # to improve them. I suspect that means we actually do better when
+ # the viewer fails to find our packaged libfontconfig.so*, falling
+ # back on the system one instead -- but diagnosing and fixing that
+ # is a bit out of scope for the present project. Meanwhile, this
+ # particular wildcard specification gets us exactly what the
+ # previous call did, without having to explicitly state the
+ # version number.
+ self.path("libfontconfig.so.*.*")
self.path("libtcmalloc.so*") #formerly called google perf tools
try:
self.path("libfmod-3.75.so")