Closed SJaffa closed 1 year ago
boost/1.78.0: WARN: Boost component 'fiber' is missing libraries. Try building boost with '-o boost:without_fiber'. (Option is not guaranteed to exist) ERROR: boost/1.78.0: Error in package_info() method, line 1592 raise ConanException("These libraries were expected to be built, but were not built: {}".format(non_built)) ConanException: These libraries were expected to be built, but were not built: {'boost_fiber'}
We need to close https://github.com/astro-informatics/sopt/issues/279 before we can make progress with this
The algo_factory that creates the ImagingForwardBackward
object in purify is templated by the algorithm type https://github.com/astro-informatics/purify/blob/5833dbd35f214212aa17c368984bcae180cf5dee/cpp/main.cc#L440 but the template parameter is checked by an if statement and there's a different branch for each algorithm class https://github.com/astro-informatics/purify/blob/5833dbd35f214212aa17c368984bcae180cf5dee/cpp/purify/algorithm_factory.h#L142
:zany_face:
The Cmake flags for building SOPT are specified in cmakefiles/LookUpSopt.cmake so we can add an option in the conanfile.py
to turn cppflow
on and off and pass that through to SOPT like they do with openmp
and logging
TODO: Add these instructions to README
To build purify
using this PR and https://github.com/astro-informatics/sopt/pull/350
git clone git@github.com:UCL/cppflow.git
conan create ./cppflow/ -pr:h=default -pr:b=default
git clone git@github.com:astro-informatics/sopt.git
conan create ./sopt/ --build missing -s compiler.libcxx=libstdc++11 -o cppflow=on -pr:h=default -pr:b=default
conan install .. --build missing -o cppflow=on -pr:h=default -pr:b=default
conan build ..
Looking at things here I think that this is up to date now with what I expect, keeping the forward-backward tensorflow test out of it for a separate PR. 👍
Ok, from my point of view the only outstanding item is updating the documentation workflow. I think everything else looks good to go.
I've checked the installation instructions in the readme work for me from a clean setup, I've updated the documentation workflow (although not tested this as it only runs when we push to development branch) and the other couple of outdated bits in the readme I've made new issues as they can be solved separately. So if someone could check over the docs workflow I think this is finally ready to merge! (assuming the tests pass...)
Starting from the development branch, first issue was the same spdlog error as in sopt
cppflow==on
option is passed to cmakeclass GProximal
to thesopt::algorithm
namespace