Activity for Dan Fandrich
-
Dan Fandrich created ticket #402
Typo in petcat BASIC list
-
That solves the test problem! Thanks. Looking forward to the next version bump and seeing version 2025.2 soon☺
-
Applying the rediffed https://sourceforge.net/p/optipng/bugs/90/#685a patch to 0.7.9 fixes the tests for me!
-
Is the issue in the test or in the underlying code? Is it safe to just ignore the test failure?
-
Test failure Assertion `num_trans == 1' failed.
-
Compile error on wildcard.c on Linux with GNU make
-
Transparent decompression in libxml2 is deprecated
-
Support using system libdeflate [patch]
-
I've created https://github.com/Enselic/recordmydesktop/pull/41
-
Compile failure with GNU libstdc++ ver. 12.0.1
-
Here's a patch that fixes the problem for me.
-
warning: pointer may be used after ‘realloc’
-
Compiler failure on gcc 12 due to redefinition of __unused__
-
gcc 12 error shows potential bug
-
I was able to build 1.0.8 on Python 3.8 with "make" after applying the following patch. I haven't actually run this version to see that this patch is correct or that it even works, though, nor tried to get the test suites running: diff -ru /tmp/pam-python-1.0.8/src/pam_python.c /home/dan/chroot-cauldron/tmp/pam-python-1.0.8/src/pam_python.c --- /tmp/pam-python-1.0.8/src/pam_python.c 2020-11-10 01:58:10.000000000 -0800 +++ /home/dan/chroot-cauldron/tmp/pam-python-1.0.8/src/pam_python.c 2021-01-18...
-
Moved to Github
-
multiple definitions error
-
libexif-0.6.21 Heap Buffer Overflow due to a programming mistake
-
This is fixed in 0.6.22
-
Use-after-free in hmac_test.c
-
To be explicit, this is a feature request for automatically using pam_wrapper for tests if autodetected or requested by the user.
-
Running tests without root
-
It probably was a huge, hard job :-)
-
I used the Fedora PyPAM (thanks for sending that pointer OOB) and got some test failures. The first is a problem in the test harness itself: File "test.py", line 64, in assert_results assert len(expected_results) == len(results), (i, expected_results[len(results)]) UnboundLocalError: local variable 'i' referenced before assignment This is because "results" was empty so the for loop never got a chance to set i. After working around that, I realized my patch in the previous comment was insufficient...
-
FWIW, I tried the tree at https://sourceforge.net/u/anders_blomdell/pam-python/ci/f7a9e21bde726b8d68815d1e9572a0636d488a5c/tree/ with Python 3.8.2 and pam 1.3.1 on Linux x86_64 and it seems to work (it passes ctest, anyway). The install target doesn't work because it's missing the same glob in pam_python3.\*.so that the previous ln -s command has (it's missing in "cp build/lib.*/pam_python3.so $(DESTDIR)$(LIBDIR)). I couldn't find a PyPAM that works with Python 3 to try test.py but ctest passes,...
-
master as of now compiles fine for me with binutils 2.34 and 2.33.
-
Thanks! I just moments ago figured out the issue as well and came up with an essentially identical patch, so I'm glad I was on the right track.
-
To save trawling through the logs, this is on Linux using g++ 9.3.0 using libstdc++ 9.3.0 and glibc 2.31.
-
Test_Namebase.sort_order test fails on ARM
-
Python 3 patch
-
This patch fixes the build for me using binutils 2.34. It will cause compile failures for previous versions, so it needs to be made conditional if compatibility with older versions is needed.
-
Splitting on Artist/Title results in corrupted titles
-
Heap Out-of-Bounds Memory Access
-
This was fixed in commit d0ec90fd.
-
Timeout (78443587)
-
The fix mentioned in the previous comment was submitted as commit 6aa11df5.
-
Timeout (37792047)
-
The fix mentioned in the previous comment was submitted as commit 6aa11df5/
-
exif crashes when removing thumbnail from jpeg image
-
I'm attaching a file that seems to show the issue here, although it doesn't give me a SIGSEGV (yet) but does show a lot of Conditional jump or move depends on uninitialised value(s) errors under valgrind. The underlying issue was solved in commit 72e9fb36 by ensuring the size parameter is initialized to 0.
-
Here is the proposed patch in unified diff form. Rather than 0xffff & does it work with an explicit cast to the enum type? And is that change really necessary on entry->components since both sides are unsigned?
-
exif --xml-output should escape field values
-
This is fixed in commit c62b9a73. Thanks for reporting it!
-
EXIF not read if Photoshop IRB Marker is present
-
The problem was solved in a more general way in commit a774b0d4.
-
Integer overflow in libexif/canon/exif-mnote-data-canon.c
-
Command line exif program removes XMP information in generated JPEGs.
-
The attached image doesn't have an XMP section but I was able to easily reproduce the problem. The issue is probably that XMP and EXIF are both stored in an APP1 but exif only cares about EXIF. I agree, it should preserve any sections it doesn't recognize, including APP1 sections that are not EXIF.
-
The library lags behind from the exif standard.
-
Shutterspeeds are rounded to nearby rational number
-
That example images contains no EXIF tags, so I can't see what you mean. The only code I see that touches an existing EXIF_TAG_EXPOSURE_TIME tag is exif_entry_fix() and that will only convert a signed rational into a directly equivalent unsigned rational; it won't scale the values at all. You might be talking about exif_entry_get_value(), which will round up values >=1 to the nearest second and those <1 second to the nearest 1/N. I agree, this is suboptimal, especially for values 0.1<N<10.
-
Heap Out-of-Bounds Memory Access
-
Timeout (37792047)
-
Timeout (78443587)
-
Out-of-Bounds Read - exif_data_save_data_entry
-
exif_tag_table_count fails to link with non-gcc compilers
-
Fixed in 80ed9dff. Thanks for the report!
-
-
libexif v0.6.21: fails to build with VS2008 (patch attached)
-
Duplicate of https://sourceforge.net/p/libexif/bugs/117/
-
libexif-0.6.21 Heap Buffer Overflow due to a programming mistake
-
[exif] fails to remove all tags
-
Pushed in c7fdec3c. Thanks for the patch!
-
I have a fix for this issue queued for release. I'm planning on making a release within 2 weeks to fix this and several other security issues. Would it be possible to delay disclosure of this for a short time to give time for the release?
-
Heap Out-of-Bounds Memory Access
-
Timeout (37792047)
-
Timeout (78443587)
-
The problem has been further mitigated (a further reduction in the maximum recursive parsing depth) queued for submission as a fix for issue #134. It would be great to announce this one simultaneously with #134 since it's effectively the same problem.
-
A fix for this issue is queued up for the next release.
-
This problem has been addressed in commit 5d28011c40ec86cf52cffad541093d37c263898a that is publically available.
-
I'm looking at a similar issue now. I hope to have the time to investigate this by next week.
-
A patch (unified diff format is best) is preferred over posting the complete modified file since it's easier to see what has changed, especially if the original file has changed compared to the version you originally modified. But free free to attach the entire file if you can't easily make a patch.
-
libexif hasn't had a release lately because there haven't patches submitted recently, presumably because no-one misses those features. Like most Open Source projects, libexif is maintained by volunteers and driven by the needs of its users as expressed through bug reports and patches. If you have need for better support for these tags or find their handling buggy, feel free to submit a patch or open a bug report with details!
-
Typo in message text
-
Applied—thanks!
-
Inline some small, often-used functions (closes...
-
Set text_flowed after send-hook has been applie...
-
Declare many structures const (closes #3552)
-
This command works for me with 7.34.0. If you're seeing Content-Length: 0 then the...
-
I would consider this issue Working As Intended. IPv6 link local addresses with scope...
-
If your target is using glibc and this is caused by a segfault, you can run your...
-
Curl into an infinite loop when Curl access one link
-
This site has a redirect loop. To guard against this kind of thing, you need to set...
-
libcurl is designed to be built and installed with one architecture per install path....
-
curl-config --feature reports SSL support, even if not supported
-
The "SSL" feature stands for "SSL" not "OpenSSL". GnuTLS provides SSL support to...
-
What is the curl --version output of the working version? I suspect it's not using...
-
The MD5 hash from your known_hosts file matches what curl gives, so that's not the...
-
Also, what's the output of "ssh-keygen -l -f ~/.ssh/known_hosts | grep 10.10.16.96"...
-
I don't see anything obviously wrong with your configuration. What is the output...
-
Please don't quote the entire bug when you reply--it makes it hard to see what new...
-
It's not SSL in this case but SSH, and -k disables the host fingerprint checking...
1 >
