Package: vlfeat / 0.9.21+dfsg0-3

Metadata

Package Version Patches format
vlfeat 0.9.21+dfsg0-3 3.0 (quilt)

Patch series

view the series file
Patch File delta Description
0001 Fixed a typo in the Makefile using default CPPFLAGS.patch | (download)

Makefile | 2 1 + 1 - 0 !
1 file changed, 1 insertion(+), 1 deletion(-)

 fixed a typo in the makefile, using default cppflags


0002 removed bogus commas in the Makefile.patch | (download)

make/bin.mak | 4 2 + 2 - 0 !
make/dist.mak | 8 4 + 4 - 0 !
make/dll.mak | 2 1 + 1 - 0 !
make/doc.mak | 2 1 + 1 - 0 !
make/matlab.mak | 8 4 + 4 - 0 !
make/octave.mak | 2 1 + 1 - 0 !
6 files changed, 13 insertions(+), 13 deletions(-)

 removed bogus commas in the makefile


0003 the DSO now has an SONAME.patch | (download)

make/dll.mak | 2 2 + 0 - 0 !
1 file changed, 2 insertions(+)

 the dso now has an soname


0004 removed unneeded RPATH from the .mex files.patch | (download)

make/octave.mak | 10 0 + 10 - 0 !
1 file changed, 10 deletions(-)

 removed unneeded rpath from the .mex files


0005 octave make clean now works properly.patch | (download)

make/octave.mak | 8 6 + 2 - 0 !
1 file changed, 6 insertions(+), 2 deletions(-)

 octave 'make clean' now works properly

Two fixes:

1. 'make clean' now always cleans up octave stuff, even if octave builds are
   off. This works to clean up from previous runs
2. octave now cleans up its .o files

0006 the mex .d file generator now works. It has include .patch | (download)

make/octave.mak | 5 3 + 2 - 0 !
1 file changed, 3 insertions(+), 2 deletions(-)

 the mex .d file generator now works. it has include-path issues


0007 removed architecture detection logic.patch | (download)

Makefile | 64 4 + 60 - 0 !
make/bin.mak | 18 0 + 18 - 0 !
make/dll.mak | 22 0 + 22 - 0 !
make/matlab.mak | 62 0 + 62 - 0 !
make/octave.mak | 8 0 + 8 - 0 !
5 files changed, 4 insertions(+), 170 deletions(-)

 removed architecture detection logic

This isn't needed on debian, and builds were failing on i386 machines since the
detection was seeing a 64-bit system for some reason

0008 Removed google adsense and google analytics from the.patch | (download)

docsrc/vlfeat-website-template.xml | 37 0 + 37 - 0 !
1 file changed, 37 deletions(-)

 removed google adsense and google analytics from the docs

This violated the user's privacy:

 http://lintian.debian.org/tags/privacy-breach-google-adsense.html

0009 I disable SSE2 and AVX to work on Debian s wide arra.patch | (download)

Makefile | 4 2 + 2 - 0 !
1 file changed, 2 insertions(+), 2 deletions(-)

 i disable sse2 and avx to work on debian's wide array of arches


0011 Handling new target of mkoctfile M . Closes 756222.patch | (download)

make/octave.mak | 2 1 + 1 - 0 !
1 file changed, 1 insertion(+), 1 deletion(-)

 handling new target of 'mkoctfile -m'. (closes: #756222)


0012 octave mex compiler now looks for headers in.patch | (download)

make/octave.mak | 2 1 + 1 - 0 !
1 file changed, 1 insertion(+), 1 deletion(-)

 octave mex compiler now looks for headers in .

This was required to build some files:

  "/usr/bin/mkoctfile" \
      -Itoolbox -M "./toolbox/imop/vl_imsmooth.c"
  In file included from ./toolbox/imop/vl_imsmooth.c:164:0:
  ./toolbox/imop/vl_imsmooth.c:17:23: fatal error: vl/float.th: No such file or directory
   #include <vl/float.th>
                         ^
  compilation terminated.

0013 removed stray html tag.patch | (download)

docsrc/tutorials/encode.html | 2 1 + 1 - 0 !
1 file changed, 1 insertion(+), 1 deletion(-)

 removed stray html tag


0013 I don t try to define mwSize and mwIndex if using Oc.patch | (download)

toolbox/mexutils.h | 2 1 + 1 - 0 !
1 file changed, 1 insertion(+), 1 deletion(-)

 i don't try to define mwsize and mwindex if using octave

Octave already defines these

0014 Removed unconditional toolbox doc matlab helpsearch .patch | (download)

make/doc.mak | 1 0 + 1 - 0 !
1 file changed, 1 deletion(-)

 removed unconditional toolbox/doc/matlab/helpsearch-v3/segments.gen
 build rule

We already have a conditional one to build this only when using matlab

0015 Using the stock getopt.patch | (download)

src/mser.c | 2 1 + 1 - 0 !
src/test_getopt_long.c | 85 0 + 85 - 0 !
vl/generic.c | 1 0 + 1 - 0 !
vl/getopt_long.c | 326 0 + 326 - 0 !
vl/getopt_long.h | 43 0 + 43 - 0 !
5 files changed, 1 insertion(+), 456 deletions(-)

 using the stock getopt


0016 Using the local mathjax.patch | (download)

docsrc/vlfeat-website-template.xml | 2 1 + 1 - 0 !
1 file changed, 1 insertion(+), 1 deletion(-)

 using the local mathjax


0017 Revert host.h Linux GCC require old GLIBC version on.patch | (download)

vl/host.h | 7 1 + 6 - 0 !
1 file changed, 1 insertion(+), 6 deletions(-)

 revert "host.h: linux/gcc: require old glibc version only"

This reverts commit ef0f2fd26657c31e1b7f86069a3da5178fab59a4.

This was breaking builds on non-x86 platforms. Upstream says (in an email) that
this was intended to improve backwards compatibility in some way. As a distro,
we handle that ourselves