Closed eigendude closed 2 years ago
Merging #601 (acae8ac) into master (03fd9bd) will not change coverage. The diff coverage is
n/a
.
@@ Coverage Diff @@
## master #601 +/- ##
=======================================
Coverage 46.02% 46.02%
=======================================
Files 715 715
Lines 32328 32328
Branches 17885 17885
=======================================
Hits 14880 14880
Misses 13102 13102
Partials 4346 4346
Continue to review full report at Codecov.
Legend - Click here to learn more
Δ = absolute <relative> (impact)
,ø = not affected
,? = missing data
Powered by Codecov. Last update 03fd9bd...acae8ac. Read the comment docs.
@eigendude, thanks for your interest and cooperation.
@Alexey-N-Chernyshov thanks for accepting any of the patches. I get crazy ideas like bringing Filecoin to the living room and start to make them. Anything accepted means I don't have to do it twice 🙃
Description of the Change
Now that https://github.com/filecoin-project/cpp-filecoin/pull/128 is merged, I've got Fuhon building without hunter again.
The approach is to delegate finding dependencies to the build system using the typical CMake approach, using find modules. I've included three find modules in this PR.
I only included find modules for three of the dependencies. The reason is because downstream, I just use the hunterized versions of dependencies. However, the hunter team made the unfortunate but easy choice of forking repos. This means that many dependencies have fallen behind the versions used by my team's app. We can't downgrade three of the packages, so I wrote CMake find modules for these three dependencies.
The final patch, for filecoin_ffi, also means that git submodules are no longer needed. This is advantageous, because git submodules cause problems for a fair number of build systems and packagers. And who hasn't forgotten
--recursive
.It's an open question if we want CMake find modules for any additional dependencies. I could possibly add this to my List Of Things To Do (tm).
Benefits
Possible Drawbacks
CMake code is added, currently totaling about 70 lines. Normally any lines added increases the load of maintenance, but code that doesn't ship has much less of an impact.
Alternate Designs [optional]
I currently use an alternative approach for the Curl dependency - I just patch out the
CONFIG
parameter, and ship the patch with the rest of my work. The exact patch is:By not including Curl in this PR, it increases my downstream technical debt, but not significantly.