Repository navigation
dev.hunter.root
As you probably have noticed there is no need to clone this repository to use Hunter since all packages will be downloaded automatically by HunterGate module. The fact that URL/SHA1 arguments used by HunterGate can be checked by finding next lines in log:
-- [hunter] Initializing Hunter workspace (a98628627c4f4c3ce11ac250fde31a7dbd264a15)
-- [hunter] /.../hunter.tar.gz
-- [hunter] -> /.../_Base/Download/Hunter/.../a986286
...
-- [hunter] [ Hunter-ID: a986286 | Config-ID: 496782c | Toolchain-ID: 5a219cb ]
Where hunter-id is the first 7 digits of SHA1 of archive (i.e. first 7 digits of SHA1 argument).
However for development purposes like adding new package there is git-based functionality in HunterGate. Do clone and set HUNTER_ROOT environment variable:
> git clone https://github.com/ruslo/hunter
> export HUNTER_ROOT="`pwd`/hunter"In this case URL/SHA1 of HunterGate will not be used!
Check logs:
-- [hunter] [ Hunter-ID: xxxxxxx | Config-ID: 496782c | Toolchain-ID: 5a219cb ]
Hence for example you need to do git pull to download new changes since modifying HunterGate arguments will not have any effects for you (but will have for non-git users, so may be needed too). There are other peculiarities with such approach so probably it make sense to not mix "user" and "development" versions of Hunter. You can achieve this by setting HUNTER_ROOT environment variable to an empty directory, do the clone to another one and use jenkins.py script for testing (see section below).
Using git-based version may be quite tricky for the beginners, have some pitfalls and require basic understanding of how Hunter works. Here are some scenarios:
- Add package
Foowith version1.2.3to Hunter modules - Build package
Foo(DONEstamp saved in Hunter working directory_Base) - Change
URLofFoopackage but keep version the same (i.e.1.2.3) - Run build -> Hunter see
DONEstamp for version1.2.3and doesn't run rebuild! - Change both
URLand version of package Foo (e.g.1.2.4) - Run build -> Hunter don't see
DONEstamp for version1.2.4, build Foo and save newDONEstamp
- Build cacheable package
Foo-> Hunter saveDONEstamp and*.tar.gzarchive with packed installated files - Change internal files like schemes or other CMake modules which should update build instructions of
Foo - Run build -> Hunter see
DONEstamp forFooand will do nothing - Remove
DONEstamp - Run build -> Hunter don't see
DONEstamp but found*.tar.gzin cache so will just unpack archive without triggering build (your updates not applied!) - Remove
DONEstamp and metadata from Cache - Run build -> Hunter will rebuild
Foowith new instructions
To avoid such problems script jenkins.py can be used. This script will pack current Hunter directory and run build instructions for project specified by PROJECT_DIR environment variable. Note that this script will create his own HUNTER_ROOT directory (i.e. environment variable HUNTER_ROOT will not be used):
# check you have python3
> python3 --version
Python 3.4.3
# download build.py script
> git clone https://github.com/ruslo/polly
> export PATH="`pwd`/polly/bin:$PATH"
> which build.py
/.../polly/bin/build.py
# clone Hunter repository
> git clone https://github.com/ruslo/hunter
> cd hunter
# run build for some package, like GTest
# environment variable TOOLCHAIN will be used as an argument for `build.py --toolchain`
[hunter]> TOOLCHAIN=default PROJECT_DIR=examples/GTest ./jenkins.py --verbose
# script will pack current archive in `_testing` directory
-- [hunter] Initializing Hunter workspace (f0550216489047a310a27c2b5ac95c70e7e878bf)
-- [hunter] /.../hunter/_testing/hunter.tar.gz
-- [hunter] -> /.../_testing/Hunter/_Base/Download/Hunter/unknown/f055021About build.py and toolchains: link
See jenkins.py --help for more options:
-
--nocreatewill not pack new Hunter archive but reuse old one (_testing/hunter.tar.gz). This may be useful for testing several packages one-by-one. Also it means that any new changes in Hunter internal modules will not be used. -
--clearremove all testing directories, i.e._testingdirectory -
--clear-except-downloadremove_Base/<hunter-id>and_Base/Cachedirectories but keep_Base/Download. This may be helpful for doing cleanup but will save archives with sources for future testing. Note that there is no need to enter full option name, i.e../jenkins.py --clear-exceptwill work fine too
Feel free to open new issue if you want to ask any questions.
- HOME
- Packages
- Android-Apk
- Android-Modules
- Android-SDK
- ArrayFire
- Assimp
- Async++
- Avahi
- BZip2
- Boost
- Boost.compute
- Boost.process
- CLAPACK
- Comet
- CppNetlib.URI
- CsvParserCPlusPlus
- Eigen
- Expat
- GPUImage
- GSL
- GTest
- GMock
- Igloo
- IntSizeof
- Jpeg
- JsonSpirit
- LLVM (Clang)
- Leathers
- Libssh2
- Lua
- MySQL-client
- nlohmann::json
- OpenCV
- OpenSSL
- PNG
- Poco C++
- PostgreSQL
- Protobuf
- Qt
- Qt-Android-CMake
- QtQmlManager
- RapidJSON
- Sugar
- TIFF
- Washer
- WTL
- ZeroMQ
- ZMQPP
- ZLIB
- rabbitmq-c
- sparsehash
- ccv
- cereal
- convertutf
- crashpad
- cvmatio
- dest
- dlib
- eos
- flatbuffers
- fmt
- freetype
- glm
- glog
- half
- hdf5
- ios_sim
- irrXML
- libyuv
- log4cplus
- minizip
- odb-pgsql
- odb-mysql
- ogles_gpgpu
- openddlparser
- poly2tri
- polyclipping
- spdlog
- szip
- thread-pool-cpp
- tinydir
- websocketpp
- msgpack
- wxWidgets
- xgboost
Don't see packages you need? Feel free to leave a package request.