-
test: Really fix mirax-slidedat-* errors on glib >= 2.51.0
Using Unicode characters in the regex produces a regex which fails to match the output.
bgilbert committedApr 24, 2017
-
test: Fix mirax-slidedat-* errors on glib >= 2.51.0
glib 10c490cdfe started using Unicode quotes in error strings. We can't match these with "." because they're multi-byte sequences and the error output is in a Python 2 str.
bgilbert committedApr 19, 2017
-
test/driver: Fix cross-testing on recent Cygwin
Fixes error: Traceback (most recent call last): File "driver", line 70, in <module> _cygwin = ctypes.CDLL('cygwin1', use_errno=True) File "/usr/lib/python2.7/ctypes/__init__.py", line 365, in __init__ self._handle = _dlopen(self._name, mode) OSError: No such file or directorybgilbert committedSep 4, 2016
-
configure.ac: Build Windows resources iff building for Windows host
Don't assume we need them just because windres is available. Do assume that windres is available when building for Windows. https://lists.andrew.cmu.edu/pipermail/openslide-users/2016-July/001274.html
bgilbert committedJul 30, 2016
-
bgilbert committed
Jul 6, 2016 -
Doxyfile: Update for Doxygen 1.8.11
bgilbert committedJul 6, 2016
-
Doxyfile: Update for Doxygen 1.8.10
bgilbert committedJun 12, 2016 -
Update package-wide copyrights for 2016
bgilbert committedJun 12, 2016 -
bgilbert committed
Jun 12, 2016
-
configure: Ubuntu Lucid is EOL; update comment
bgilbert committedJun 10, 2016 -
configure: Try to check for libjpeg with pkg-config
libjpeg-turbo >= 1.5.0 provides a pkg-config file.
bgilbert committedJun 10, 2016
-
test: Use platform-specified stack frame alignment on Win32
The Win32 calling convention only requires 4-byte alignment of stack frames, but GCC expects 16-byte alignment. This can cause crashes when interworking with MSVC, such as openslide/openslide-winbuild#9, if OpenSlide is not built with compensatory compiler flags (such as -mstackrealign). On Win32 only, build the test programs with the stack frame alignment specified by the Win32 calling convention, in order to check that interwork is functional. This can be disabled by configuring with --disable-checking-windows-abi-compat in case interwork is not needed.
bgilbert committedJun 4, 2016
-
configure.ac: Move all compiler warning flags to WARN_CFLAGS
for consistency, and ease of overriding from the command line.
bgilbert committedJun 3, 2016 -
configure.ac: Use gl_WARN_ADD to detect -Wno-pedantic-ms-format
bgilbert committedJun 3, 2016
-
test/driver: Add way to mute tests which are expected to fail
Add OPENSLIDE_TEST_XFAIL and OPENSLIDE_VALGRIND_XFAIL environment variables, which can contain a comma-separated list of tests which are expected to fail during run and valgrind respectively. Complain if those tests end up passing, but otherwise, just warn about the expected failure and continue without affecting the exit status.
bgilbert committedMay 19, 2016
-
Use g_utf8_to_utf16() to compute Windows file paths
On glib 2.47.6 through at least 2.48.1, g_convert() adds a BOM to the beginning of the returned string, causing _wfopen() to fail. Switch to g_utf8_to_utf16(), which should also be more efficient. Fixes #176.
bgilbert committedMay 16, 2016
-
Document that g_key_file_new() should be thread-safe in glib >= 2.48.1
bgilbert committedMay 14, 2016
-
test/driver: Disable xdelta LZMA secondary compression during pack
xdelta 3.0.11 enables LZMA secondary compression by default, but distro binaries don't always support LZMA. This was breaking the test suite on Precise, MacPorts, and Cygwin.
bgilbert committedJan 25, 2016
-
Merge pull request #172 from agoode/master
Fix an integer overflow in Hamamatsu parsing Fixes #171.
bgilbert committedJan 24, 2016
-
agoode committed
Jan 23, 2016 -
Use INT32_MAX instead of INT_MAX
agoode committedJan 23, 2016 -
Fix an integer overflow in Hamamatsu parsing
We can overflow when reading a VMS file. This is bad since we index on the value later and can do an out of bounds memory access. Thanks to Kostya Kortchinsky.
agoode committedJan 23, 2016
-
hamamatsu: Use sizeof rather than hardcoding buffer sizes
bgilbert committedJan 7, 2016 -
hamamatsu: Rename extract_one_optimisation()
It actually extracts one JPEG worth of optimisations.
bgilbert committedJan 7, 2016 -
misc: Add scripts to report VMS tile offsets
bgilbert committedJan 7, 2016 -
hamamatsu: Use local variable to improve readability
bgilbert committedJan 7, 2016
-
hamamatsu: Break hamamatsu_vms_part2() loop in half
Ignore the map file in the second loop.
bgilbert committedJan 7, 2016 -
hamamatsu: Drop unnecessary variable
bgilbert committedJan 7, 2016 -
hamamatsu: Don't try to read optimisations for VMS map file
None of my sample VMS files have optimisation data for the map file. On slides with a single Z-layer this would cause us to harmlessly give up on the .opt file, but slides with multiple Z-layers place the additional optimisation data after the optimisations for Z-layer 0. This was causing restart marker position mismatches for those slides. Stop trying to read optimisations for the map file. Fixes #165.
bgilbert committedJan 7, 2016 -
hamamatsu: Don't shadow loop index in inner loop
bgilbert committedJan 7, 2016 -
hamamatsu: Minor cleanup to vms_part2 JPEG verification logic
bgilbert committedJan 7, 2016
-
bgilbert committed
Nov 9, 2015 -
test/driver: Add command to check for symbols with incorrect visibility
bgilbert committedNov 9, 2015