-
Notifications
You must be signed in to change notification settings - Fork 3.3k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Salmon gcc7 #10344
Salmon gcc7 #10344
Conversation
Hrmm, this seemed to work .... not so sure what's different between this and before when it failed. Going to try uncommenting OSX now too. |
Try building on OSX
Hrmmm .. looks like this icu58 issue is still popping up on OSX, but has gone away under linux on the newer compiler. I'd really like to figure this out so I can provide a mac executable through bioconda. |
recipes/salmon/build_config.yaml
Outdated
@@ -0,0 +1,6 @@ | |||
c_compiler: | |||
- gcc # [linux] | |||
- clang # [osx] |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Rob can you try to replace clang here with GCC as well?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Ok; just pushed. I didn't even know that building with GCC on OSX was an option here :).
trying GCC/GXX even on Mac
It used clang anyway, it seems .... |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Ahhh, so it wasnt even fixed on linux yet. Well, at least the error seems the same in both cases! Is this affecting any other packages?
Not really as we do not build many pkgs with the new compilers. My assumption is that we need to rebuild iuc with the new compilers before we build Salmon :( |
Can that be done here, or is that coming from conda-forge? Can we submit PRs there? |
Its a larger effort we are working on. The migration to the new compilers should happen soon. But it will take time, nothing we can influence right now. Its mainly CI build time dependent ... If someone want to buy more CI build-time - more then welcome :) I will keep you posted, sorry. |
Ok, so I get it to build locally if I force the boost from anaconda cloud and not the one from conda-forge. The conda-forge ppl aren't building a c++11 enabled boost :(. Perhaps there is a more elegant way to do this. |
revert to clang on OSX.
ahh, so no we are getting something different popping up on OSX. Perhaps this one is solvable. Any thoughts @bioconda/osx ? |
bump llvm_version
version
version again
|
Now it's back at a damn icu58 linker issue under OSX ... . But, the icu being pulled in is v58. From the build log:
why won't it link? There are also some even stranger errors:
and
etc. Perhaps this boost was not linked against |
ok, let's go all in on the new stuff and try to fix it from that direction
try different boost on osx?
this is a frustrated commit message ...
Try boost instead of libboost --- do we get the same error on OSX?
See if we can use boost instead of libboost on linux
nope ... `boost` instead of `libboost` fails under linux.
horrible, horrible, horrible ...
nope. nope, nope, nope --- we *need* `libboost` on linux for now. Why doesn't it work on OSX?
try this
update md5sum
50th time's a charm?
51st time's a charm ... I'm gonna guess not.
Holy shit! |
Ok --- last attempt *worked*! Now, try to make everything C++14, and re-enable linux build.
have to update md5sum
Note : This is currently pulling from develop. I will be cutting a new release (tagged) imminently. Please don't merge this yet. |
try and build tagged v0.11.3
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Holy cow! Amazing @rob-p. Thanks a lot for all the effort - very appreciated.
host: | ||
- boost | ||
- icu | ||
- libboost # [linux] |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
the selector is now not needed here ...
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thanks @bgruening --- yes, I agree that this is not needed (and probably the same with the libcxx below). Should I try to bump the version and remove these and see if it compiles?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
That would be great, yes. Thanks.
- boost | ||
- icu | ||
- libcxx | ||
- libboost # [linux] |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
same here
- {{ compiler('cxx') }} | ||
- {{ compiler('c') }} | ||
#- clangdev =={{ llvm_version }} # [osx] | ||
#- libcxx =={{ llvm_version }} # [osx] | ||
- libcxx |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
is this needed?
This is a test to see if we can build salmon using the new compiler (gcc7) ... at least on linux. This works locally using the "bootstrap" method, so let's see what happens on Circle CI.