| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
| |
Package-Manager: Portage-2.3.62, Repoman-2.3.12
Signed-off-by: Horea Christian <chr@chymera.eu>
|
|
|
|
|
| |
Package-Manager: Portage-2.3.62, Repoman-2.3.12
Signed-off-by: Martin Mokrejs <mmokrejs@fold.natur.cuni.cz>
|
|
|
|
|
|
|
|
|
|
| |
RDEPEND uses the = operator with no revision:
'=sci-libs/tensorflow-1.0.1';
if any revision is acceptable, use '~' instead;
if only -r0 then please append '-r0' to the dep
Package-Manager: Portage-2.3.62, Repoman-2.3.12
Signed-off-by: Martin Mokrejs <mmokrejs@fold.natur.cuni.cz>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Non-maintainer commit fixing:
RDEPEND.suspect 2
sci-mathematics/dolfin/dolfin-2016.2.0-r2.ebuild: 'dev-lang/swig:0='
sci-mathematics/dolfin/dolfin-2017.1.0.ebuild: 'dev-lang/swig:0='
All swig ebuilds define only SLOT="0"
Package-Manager: Portage-2.3.62, Repoman-2.3.12
Signed-off-by: Martin Mokrejs <mmokrejs@fold.natur.cuni.cz>
|
|
|
|
|
|
|
| |
Closes: https://bugs.gentoo.org/671510
Package-Manager: Portage-2.3.62, Repoman-2.3.12
Signed-off-by: Jian Cao <sworden.cao@gmail.com>
Signed-off-by: Justin Bronder <jsbronder@gentoo.org>
|
|
|
|
|
| |
Package-Manager: Portage-2.3.62, Repoman-2.3.12
Signed-off-by: Horea Christian <horea.christ@yandex.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Add more regexps specifying which parts of the test suite should
not be compiled at all.
We have so far problems with say ncbi-tools++-18.0.0 libs installed
in ${EPREFIX}/usr/lib64/ncbi-tools++/ which get picked during linking
of a currently emerged package, like ncbi-tools++-22.0.0. That happens
due to LDPATH being set through
```
echo "LDPATH=${EPREFIX}/usr/$(get_libdir)/${PN}" > ${S}/99${PN}
doenvd "${S}/99${PN}"
```
which results in
```
$ cat /etc/env.d/99ncbi-tools++
LDPATH=/usr/lib64/ncbi-tools++
$
```
and later used during linking of a future version. Some symbols will be
resolved by ${EPREFIX}/usr/lib64/ncbi-tools++/ from older version,
some probably not. It is confusing to debug the errors popping up
during compilation/linking.
Aaron Ucko from NCBI proposed:
To avoid interference from old installations, please try passing
`LDFLAGS=-Wl,-rpath-link,"${S}"_build/lib` to configure, no patches
(likely) needed.
@kiwifb François Bissey commented with:
<quote>
LDPATH is used to generate the ld.so.conf.d files. So it is important
to have it to be able to find the libraries that are not in a standard
path. LDPATH is a Gentoo special. You need to regenerate the cache with
ldconfig after installing such a library and such a path.
...
$ /usr/bin/x86_64-pc-linux-gnu-g++ -std=gnu++11 -Wl,--verbose
-Wl,-rpath,/usr/lib/gcc/x86_64-pc-linux-gnu/8.3.0
-Wl,--enable-new-dtags -Wl,-export-dynamic -pthread -Wl,-O1
-Wl,--as-needed -g speedtest.o
-L/scratch/var/tmp/portage/sci-biology/ncbi-tools++-22.0.0/work/\
ncbi_cxx--22_0_0_build/lib
-lprosplign -lxalgoalignutil -ltaxon1 -lxalgoseq -lxcleanup
-lxobjedit -ltaxon3 -lvalid -lvalerr -lproteinkmer -lxblast
-lxalgoblastdbindex -lcomposition_adjustment -lxalgodustmask
-lxalgowinmask -lseqmasks_io -lseqdb -lblast_services -lxalnmgr
-lxobjutil -lxobjread -lvariation -lcreaders -lsubmit -lxnetblastcli
-lxnetblast -lblastdb -lscoremat -ltables -lxqueryparse -lxregexp
-lncbi_xloader_genbank -lncbi_xreader_id1 -lncbi_xreader_id2
-lncbi_xreader_cache -ldbapi_driver -lncbi_xreader -lxconnect
-lid1 -lid2 -lxobjmgr -lgenome_collection -lseqedit -lseqsplit
-lsubmit -lseqset -lseq -lseqcode -lsequtil -lpub -lmedline
-lbiblio -lgeneral -lxser -lxutil -lxncbi -lxcompress -lz -lbz2
-llzo2 -lz -lnsl -ldl -lpcre -llmdb -lpthread -ldw -ldl -lm
-lpthread -o speedtest
It cannot find libmlacli.so. Note that there isn’t a `-lmlacli`
anywhere in the above line. It is searched for because you are
compiling an executable and executable cannot have unresolved
symbols. It must in the table of needed libraries for libxobjedit.so
libmlacli.so needed by /scratch/var/tmp/portage/sci-biology/\
ncbi-tools++-22.0.0/work/ncbi_cxx--22_0_0_build/lib/libxobjedit.so
Because there is no path or possibly a final path in libxobjedit.so
that library cannot be found.
Two solutions here:
1) add -lmlacli in the compilation line above
2) Precede the whole line with
LD_LIBRARY_PATH=/scratch/var/tmp/portage/sci-biology/\
ncbi-tools++-22.0.0/work/ncbi_cxx--22_0_0_build/lib
The LDPATH setting probably enabled you to find a previously
installed version of the library.
François
PS: You can pass that analysis upstream because -lmlacli should
probably be added by right.
</quote>
Later François also wrote:
<quote>
rpath are usually the preferred solution but it is not the end all
solution. Looking for LDPATH in your /etc/env.d/ may shock you.
In the case of ncbi, adding path is only one of the solution I
would consider. I don’t see why they need to install in
/usr/lib{,64}/ncbi-tools++ rather than just /usr/lib{,64} in
the first place. It is not even versioned to allow several versions
installed in parallel. It is not plugins or “modules” or other
special kinds of shared objects.
Possibly it is some language runtime - at best. Does it use names
that could collide easily with something else? May be some of them
should be prefixed then.
</quote>
make[3]: Entering directory '/scratch/var/tmp/portage/sci-biology/\
ncbi-tools++-22.0.0/work/ncbi_cxx--22_0_0_build/build/app/speedtest'
Updating dependency information for /scratch/var/tmp/portage/sci-biology/\
ncbi-tools++-22.0.0/work/ncbi_cxx--22_0_0/src/app/speedtest/speedtest.cpp.
/usr/bin/x86_64-pc-linux-gnu-g++ -std=gnu++11 -c -Wall -Wno-format-y2k
-pthread -pipe -mpclmul -mpopcnt -march=native -ftree-vectorize -fPIC
-g -D_DEBUG -D_LARGEFILE_SOURCE -D_FILE_OFFSET_BITS=64 -D_LARGEFILE64_SOURCE
-D_MT -D_REENTRANT -D_THREAD_SAFE
-I/scratch/var/tmp/portage/sci-biology/ncbi-tools++-22.0.0/work/\
ncbi_cxx--22_0_0_build/inc
-I/scratch/var/tmp/portage/sci-biology/ncbi-tools++-22.0.0/work/\
ncbi_cxx--22_0_0/include
/scratch/var/tmp/portage/sci-biology/ncbi-tools++-22.0.0/work/\
ncbi_cxx--22_0_0/src/app/speedtest/speedtest.cpp
-o speedtest.o
/usr/bin/x86_64-pc-linux-gnu-g++ -std=gnu++11
-Wl,-rpath,/usr/lib/gcc/x86_64-pc-linux-gnu/8.3.0
-Wl,--enable-new-dtags -Wl,-export-dynamic -pthread -Wl,-O1
-Wl,--as-needed -g speedtest.o
-L/scratch/var/tmp/portage/sci-biology/ncbi-tools++-22.0.0/work/\
ncbi_cxx--22_0_0_build/lib
-lprosplign -lxalgoalignutil -ltaxon1 -lxalgoseq -lxcleanup -lxobjedit
-ltaxon3 -lvalid -lvalerr -lproteinkmer -lxblast -lxalgoblastdbindex
-lcomposition_adjustment -lxalgodustmask -lxalgowinmask -lseqmasks_io
-lseqdb -lblast_services -lxalnmgr -lxobjutil -lxobjread -lvariation
-lcreaders -lsubmit -lxnetblastcli -lxnetblast -lblastdb -lscoremat
-ltables -lxqueryparse -lxregexp -lncbi_xloader_genbank -lncbi_xreader_id1
-lncbi_xreader_id2 -lncbi_xreader_cache -ldbapi_driver -lncbi_xreader
-lxconnect -lid1 -lid2 -lxobjmgr -lgenome_collection -lseqedit -lseqsplit
-lsubmit -lseqset -lseq -lseqcode -lsequtil -lpub -lmedline -lbiblio
-lgeneral -lxser -lxutil -lxncbi -lxcompress -lz -lbz2 -llzo2 -lz -lnsl
-ldl -lpcre -llmdb -lpthread -ldw -ldl -lm -lpthread -o speedtest
/usr/lib/gcc/x86_64-pc-linux-gnu/8.3.0/../../../../x86_64-pc-linux-gnu/\
bin/ld: warning: libmlacli.so, needed by /scratch/var/tmp/portage/\
sci-biology/ncbi-tools++-22.0.0/work/ncbi_cxx--22_0_0_build/lib/\
libxobjedit.so, not found (try using -rpath or -rpath-link)
The problem here is that the rpath would have to be provided when
you link libxobjedit.so. I checked the log and it wasn’t done.
And you definitely don’t want to do that unless you re-link without
the rpath (or the correct final rpath) at install time. For information
cmake does exactly that.
But in the end adding a path in this case is no substitute to providing
the libraries you need. When building stuff with libtool, libtool produces
test scripts that set LD_LIBRARY_PATH properly for running the tests before
install. So the test executables don’t have to be linked with rpath to
the build folder. Or the objects that you want to ship either.
</quote>
This change will ensure `LDFLAGS=-Wl,-rpath-link,"${S}"_build/lib` is
passed to `bash ./src/build-system/configure` in `src_configure()`
and at least helps to get symbols resolved during linking although it
is probably still far from ideal.
Package-Manager: Portage-2.3.62, Repoman-2.3.12
Signed-off-by: Martin Mokrejs <mmokrejs@fold.natur.cuni.cz>
|
|
|
|
|
| |
Package-Manager: Portage-2.3.62, Repoman-2.3.12
Signed-off-by: Matthias Maier <tamiko@gentoo.org>
|
|
|
|
|
| |
Package-Manager: Portage-2.3.62, Repoman-2.3.12
Signed-off-by: Horea Christian <horea.christ@yandex.com>
|
|
|
|
|
| |
Package-Manager: Portage-2.3.62, Repoman-2.3.12
Signed-off-by: Horea Christian <horea.christ@yandex.com>
|
|
|
|
|
| |
Package-Manager: Portage-2.3.62, Repoman-2.3.12
Signed-off-by: Horea Christian <horea.christ@yandex.com>
|
|
|
|
|
| |
Package-Manager: Portage-2.3.62, Repoman-2.3.12
Signed-off-by: Horea Christian <horea.christ@yandex.com>
|
|
|
|
|
|
| |
From: Torsten S.
The velvetk.pl is faulty and should not be used.
Try `kmergenie` instead, or use the SPades Assembler.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
I have the following defined due to ncbi-tools++-18.0.0 installed
on my system:
/usr/lib/gcc/x86_64-pc-linux-gnu/8.3.0/32:
/usr/lib/gcc/x86_64-pc-linux-gnu/8.3.0:
/usr/lib/gcc/x86_64-pc-linux-gnu/7.3.0/32:
/usr/lib/gcc/x86_64-pc-linux-gnu/7.3.0:
/usr/lib/llvm/7/lib32:
/usr/lib/llvm/7/lib64:
/usr/lib64/itcl4.1.1:
/usr/lib64/itk4.1.0:
/lib64:
/usr/lib64:
/usr/local/lib64:
/lib32:
/usr/lib32:
/usr/lib64/rust-1.29.2:
/usr/lib64/R/lib:
/usr/lib64/fltk:
/usr/lib64/ncbi-tools++:
/usr/lib64/staden:
grep: /etc: No such file or directory
grep: /etc/env.d: No such file or directory
/etc/env.d/99ncbi-tools++:LDPATH=/usr/lib64/ncbi-tools++
...
That in turn makes linker pick "${WORKDIR}"/ncbi_cxx--22_0_0_build/lib/*
files but later overrides some of them with /usr/lib64/ncbi-tools++/*,
depending on the ordering of -lXXXXXX arguments.
Package-Manager: Portage-2.3.62, Repoman-2.3.12
Signed-off-by: Martin Mokrejs <mmokrejs@fold.natur.cuni.cz>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
It does not link if older version is installed because linker
picks up older version of libraries and obviously does not find
newly added functions.
Upstream says:
Please try either supplying an LD_RUN_PATH environment setting
or a -Wl,-rpath-link,... flag pointing at your build tree's lib
directory. Also, please bear in mind that adding the old
installation's library directory to the central search path largely
defeats the purpose of making it a separate directory.
Package-Manager: Portage-2.3.62, Repoman-2.3.12
Signed-off-by: Martin Mokrejs <mmokrejs@fold.natur.cuni.cz>
|
|
|
|
|
| |
Package-Manager: Portage-2.3.62, Repoman-2.3.12
Signed-off-by: Martin Mokrejs <mmokrejs@fold.natur.cuni.cz>
|
|
|
|
|
| |
Package-Manager: Portage-2.3.62, Repoman-2.3.12
Signed-off-by: Martin Mokrejs <mmokrejs@fold.natur.cuni.cz>
|
|
|
|
|
| |
Package-Manager: Portage-2.3.62, Repoman-2.3.12
Signed-off-by: Martin Mokrejs <mmokrejs@fold.natur.cuni.cz>
|
|
|
|
|
| |
Package-Manager: Portage-2.3.62, Repoman-2.3.12
Signed-off-by: Martin Mokrejs <mmokrejs@fold.natur.cuni.cz>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The changes are partly based on https://github.com/gentoo/sci/pull/580
from Michael Schubert but it seems we could figure out the paths
to binaries automatically using `which`.
Block between the USE options and allow only a single to be enabled,
that is what configure requires, otherwise it enters a loop.
The hmmer is the default in configure and preferred.
Tested with `USE="hmmer -wublast -phrap -rmblast"` and other
combinations while only one is allowed.
If users have wublast/abblast/rmblast/crossmatch in the PATHS we just
pick it.
`which` returns a '.' (dot) if a binary was not found, we need to zap it
otherwise configure enters a loop.
In the end, one can get:
Add a Search Engine:
1. CrossMatch: [ Configured ]
2. RMBlast - NCBI Blast with RepeatMasker extensions: [ Configured ]
3. WUBlast/ABBlast (required by DupMasker): [ Un-configured ]
4. HMMER3.1 & DFAM: [ Configured, Default ]
5. Done
properly written into
/usr/lib64/perl5/vendor_perl/5.28.0/RepeatMaskerConfig.pm
Package-Manager: Portage-2.3.62, Repoman-2.3.12
Signed-off-by: Martin Mokrejs <mmokrejs@fold.natur.cuni.cz>
|
|
|
|
|
|
|
| |
This slipped through repoman checks.
Package-Manager: Portage-2.3.62, Repoman-2.3.12
Signed-off-by: Martin Mokrejs <mmokrejs@fold.natur.cuni.cz>
|
|
|
|
|
| |
Package-Manager: Portage-2.3.62, Repoman-2.3.12
Signed-off-by: Martin Mokrejs <mmokrejs@fold.natur.cuni.cz>
|
|
|
|
|
| |
Package-Manager: Portage-2.3.62, Repoman-2.3.12
Signed-off-by: Martin Mokrejs <mmokrejs@fold.natur.cuni.cz>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Do not require static libraries. Only bamtools are really needed.
htslib is not needed. For bam2wig samtools-0.2 series is needed.
I decided to blindly install some of the binaries created but not
copied into `./bin/`.
`src/Makefile` will gladly define mysql and sqlite together, probably
a non-sense. We force mysql as we check for it first. Not sure if
sqlite should be preferred.
I documented in the augustus-3.3.2.ebuild other "forgotten" files,
like gbrowse.conf and a template for gff2ps.
I dropped `-O3` but left in `-ggdb` in the `CXXFLAGS`.
https://github.com/Gaius-Augustus/Augustus/issues/14
https://github.com/Gaius-Augustus/Augustus/issues/50
https://github.com/Gaius-Augustus/Augustus/issues/51
https://github.com/Gaius-Augustus/Augustus/issues/52
https://github.com/Gaius-Augustus/Augustus/issues/53
Package-Manager: Portage-2.3.62, Repoman-2.3.12
Signed-off-by: Martin Mokrejs <mmokrejs@fold.natur.cuni.cz>
|
|
|
|
|
| |
Package-Manager: Portage-2.3.62, Repoman-2.3.12
Signed-off-by: Horea Christian <horea.christ@yandex.com>
|
|
|
|
|
| |
Package-Manager: Portage-2.3.62, Repoman-2.3.12
Signed-off-by: Martin Mokrejs <mmokrejs@fold.natur.cuni.cz>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Trying to get rids of FATAL errors reported by Travis.CI.
Use `if [ "${ABI}" == "amd64" ]; then`
instead of
`if use amd64; then`
Also bump to EAPI7.
Package-Manager: Portage-2.3.62, Repoman-2.3.12
Signed-off-by: Martin Mokrejs <mmokrejs@fold.natur.cuni.cz>
|
|
|
|
|
| |
Package-Manager: Portage-2.3.62, Repoman-2.3.12
Signed-off-by: Martin Mokrejs <mmokrejs@fold.natur.cuni.cz>
|
|
|
|
|
| |
Package-Manager: Portage-2.3.62, Repoman-2.3.12
Signed-off-by: Martin Mokrejs <mmokrejs@fold.natur.cuni.cz>
|
|
|
|
|
| |
Package-Manager: Portage-2.3.62, Repoman-2.3.12
Signed-off-by: Martin Mokrejs <mmokrejs@fold.natur.cuni.cz>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Do not unset contents of CFLAGS and CXXFLAGS.
Thanks to ch4rr0 at
https://github.com/BenLangmead/bowtie2/issues/239
Add deemed patch from
https://github.com/BenLangmead/bowtie2/issues/240
Added KEYWORDS.
Closes: https://bugs.gentoo.org/649824
Package-Manager: Portage-2.3.62, Repoman-2.3.12
Signed-off-by: Martin Mokrejs <mmokrejs@fold.natur.cuni.cz>
|
|
|
|
|
|
|
| |
Still does not work.
Package-Manager: Portage-2.3.62, Repoman-2.3.12
Signed-off-by: Martin Mokrejs <mmokrejs@fold.natur.cuni.cz>
|
|
|
|
|
|
|
| |
Still unfinished work.
Package-Manager: Portage-2.3.62, Repoman-2.3.12
Signed-off-by: Martin Mokrejs <mmokrejs@fold.natur.cuni.cz>
|
|
|
|
|
|
|
| |
https://github.com/gentoo/sci/issues/907#issuecomment-476420962
Package-Manager: Portage-2.3.62, Repoman-2.3.12
Signed-off-by: Martin Mokrejs <mmokrejs@fold.natur.cuni.cz>
|
|
|
|
|
|
|
|
| |
Thanks for the patch and guidance to François Bissey:
https://github.com/gentoo/sci/issues/907
Package-Manager: Portage-2.3.62, Repoman-2.3.12
Signed-off-by: Martin Mokrejs <mmokrejs@fold.natur.cuni.cz>
|
|
|
|
|
| |
Package-Manager: Portage-2.3.62, Repoman-2.3.12
Signed-off-by: Martin Mokrejs <mmokrejs@fold.natur.cuni.cz>
|
|
|
|
|
|
|
|
|
|
|
| |
We have only dev-java/htdjdk-2.3.0 in the tree as newer versions
use gradle. The tablet-1.17.08.17 lacks the lib/htsjdk-2.0.11.jar
so we cannot install it. Use sci-biology/tablet-bin instead.
https://github.com/cropgeeks/tablet/issues/6
Package-Manager: Portage-2.3.62, Repoman-2.3.12
Signed-off-by: Martin Mokrejs <mmokrejs@fold.natur.cuni.cz>
|
| |
|
|
|
|
|
| |
Package-Manager: Portage-2.3.62, Repoman-2.3.12
Signed-off-by: Martin Mokrejs <mmokrejs@fold.natur.cuni.cz>
|
|
|
|
|
| |
Package-Manager: Portage-2.3.62, Repoman-2.3.12
Signed-off-by: Martin Mokrejs <mmokrejs@fold.natur.cuni.cz>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Bump to EAPI7
Building does not work since sandbox has now no network access.
Downloading https://services.gradle.org/distributions/gradle-5.2.1-bin.zip
Exception in thread "main" java.net.SocketException: Network
is unreachable (connect failed)
htsjdk calls some `gradlew` wrapper so the recent ebuilds
need more work to use dev-java/gradle-bin package.
Package-Manager: Portage-2.3.62, Repoman-2.3.12
Signed-off-by: Martin Mokrejs <mmokrejs@fold.natur.cuni.cz>
|
| |
|
|
|
|
|
| |
Package-Manager: Portage-2.3.62, Repoman-2.3.12
Signed-off-by: Martin Mokrejs <mmokrejs@fold.natur.cuni.cz>
|
|
|
|
|
| |
Package-Manager: Portage-2.3.62, Repoman-2.3.12
Signed-off-by: Martin Mokrejs <mmokrejs@fold.natur.cuni.cz>
|
|
|
|
|
|
|
|
|
|
| |
New homepage, switch from svn to github, new license, switch
from picard to htsjdk
There is no ./classes/ so `ant build` does not work for me.
Package-Manager: Portage-2.3.62, Repoman-2.3.12
Signed-off-by: Martin Mokrejs <mmokrejs@fold.natur.cuni.cz>
|
| |
|
|
|
|
|
| |
Package-Manager: Portage-2.3.62, Repoman-2.3.12
Signed-off-by: Martin Mokrejs <mmokrejs@fold.natur.cuni.cz>
|
|
|
|
|
| |
Package-Manager: Portage-2.3.62, Repoman-2.3.12
Signed-off-by: Martin Mokrejs <mmokrejs@fold.natur.cuni.cz>
|
|
|
|
|
| |
Package-Manager: Portage-2.3.62, Repoman-2.3.12
Signed-off-by: Martin Mokrejs <mmokrejs@fold.natur.cuni.cz>
|