dev.hunter.root

Automatic download

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).

Git clone

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).

jenkins.py

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:

scenario 1

1. Add package Foo with version 1.2.3 to Hunter modules
2. Build package Foo (DONE stamp saved in Hunter working directory _Base)
3. Change URL of Foo package but keep version the same (i.e. 1.2.3)
4. Run build -> Hunter see DONE stamp for version 1.2.3 and doesn't run rebuild!
5. Change both URL and version of package Foo (e.g. 1.2.4)
6. Run build -> Hunter don't see DONE stamp for version 1.2.4, build Foo and save new DONE stamp

scenario 2

1. Build cacheable package Foo -> Hunter save DONE stamp and *.tar.gz archive with packed installated files
2. Change internal files like schemes or other CMake modules which should update build instructions of Foo
3. Run build -> Hunter see DONE stamp for Foo and will do nothing
4. Remove DONE stamp
5. Run build -> Hunter don't see DONE stamp but found *.tar.gz in cache so will just unpack archive without triggering build (your updates not applied!)
6. Remove DONE stamp and metadata from Cache
7. Run build -> Hunter will rebuild Foo with 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/f055021

About build.py and toolchains: link

See jenkins.py --help for more options:

• --nocreate will 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.
• --clear remove all testing directories, i.e. _testing directory
• --clear-except-download remove _Base/<hunter-id> and _Base/Cache directories 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-except will work fine too