Debian Patches

Status for pbseqlib/5.3.5+dfsg-13

Patch Description Author Forwarded Bugs Origin Last update
hdf5-1.10.1-support.patch cope with HDF5 C++ API changes in release 1.10.1 From HDF5 upstream commits c78f1ec8829 and b1c4fd77d9c:
.
[svn-r30309] Purpose: Fix bug HDFFR-9920
Description:
Rearranged the classes to model the relationship of HDF5 objects more
accurately. The changes included:
- moved CommonFG's methods to Group
- removed CommonFG from Group's base class list
- removed CommonFG from H5File's base class list and changed
"public H5Location" to "public Group" in the base class list
- CommonFG became unused
- result of the modified partial class diagram:
.
IdComponent
|
H5Location
/
H5Object
|
/ / \
DadaType Group DataSet
|
H5File
.
[svn-r30311] Purpose: Fix bug HDFFR-9920 cont.
Description:
Continued rearranging the classes to model the relationship of HDF5
objects more accurately. The changes included:
- in the baseclass list of Attribute, changed "public IdComponent" to
"public H5Location", because location sometime can be specified with
attribute
- moved H5A wrappers in H5Location to H5Object because H5A functions
can't be called on attribute id
- removed the stubs Attribute::iterateAttrs and Attribute::renameAttr
- removed Attribute::getFileName and Attribute::flush, because
H5Location has them
- result of the modified partial class diagram, regarding Attribute
.
IdComponent
|
H5Location AbstractDs
/ \ /
H5Object Attribute		 Gilles Filippini <pini@debian.org> no
hdf5-1.10.0-support.patch Fix build against recent Hdf5 lib Gilles Filippini <pini@debian.org> no debian 2018-07-01
meson.patch There is no point to check for gtest since its installed via Build-Depends Andreas Tille <tille@debian.org> no 2019-12-13
static+shared_lib.patch Build static and shared library Andreas Tille <tille@debian.org> no 2020-11-13
gcc-13.patch Include cstdint to fix ftbfs with gcc-13 Nilesh Patra <nilesh@riseup.net> no debian 2023-06-20
pbbam_2.3.0 Adapt to pbbam 2.3.0 / pbcopper 2.2.0 changes Michael R. Crusoe no
gcc-14.patch Potential fix for gcc-14 Nilesh Patra <nilesh@debian.org> no 2024-12-07
gcc-16.patch operator== is bool not int in C++20 which is strictly enforced in GCC 16 (The bug was filed against package pbdagcon but it was a result of this header file) Andreas Tille <tille@debian.org> no debian 2026-08-12
pbseqlib-findmaxinterval-size1-oob.patch Guard out-of-range reads when an anchor match list has exactly one entry ExhaustiveFindMaxIncreasingInterval (and the Fast-finder helper
StoreLargestIntervals) unconditionally read pos[next] where
next = cur + 1 = 1, and later pos[cur] after advancing cur to the end,
assuming at least two anchors. For a match list with a single anchor
(pos.size() == 1) this indexes past the end of the
std::vector<ChainedMatchPos>. With _GLIBCXX_ASSERTIONS
(default hardening in Debian toolchains) this aborts with
"std::vector<ChainedMatchPos>::operator[] ... Assertion '__n <
this->size()' failed", i.e. "Assertion '__n < this->size()' failed",
killing BLASR. The PBHoney 'Honey pie' tail remapping step routinely
produces single-anchor read lists, so the pbsuite autopkgtest
(Debian bug #1091721) crashes and fails on unstable/testing/stable.
Without assertions the same reads walk out of bounds (undefined
behavior), yielding garbage alignment results.
.
Fix: guard the two boundary reads (next < nPos, cur < nPos) so a
single-anchor list is processed in the existing "next - cur == 1"
single-match branch and the loop terminates cleanly. Also make
StoreLargestIntervals return early for pos.size() <= 1 (a single
anchor cannot form a cluster), which is the identical off-by-one
exposed when --fast-max-interval is used.		 Debian Med packaging team <debian-med-packaging@lists.alioth.debian.org> not-needed debian 2026-10-23

All known versions for source package 'pbseqlib'

Links