-
Notifications
You must be signed in to change notification settings - Fork 2.2k
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
Changed +fpic to +pic variant #2479
Conversation
@aprokop Are you able to fix up any of these packages? |
@citibeth If by fixing you mean a simple replacing with "{0}".format(self.compiler.pic_flag), sure, I could do that. But I won't have time to test all (or maybe even any) of them at this point. |
yes thanks that would be great
…On Dec 5, 2016 10:38 AM, "Andrey Prokopenko" ***@***.***> wrote:
@citibeth <https://github.com/citibeth> If by fixing you mean a simple
replacing with "{0}".format(self.compiler.pic_flag), sure, I could do
that. But I won't have time to test all (or maybe even any) of them at this
point.
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#2479 (comment)>, or mute
the thread
<https://github.com/notifications/unsubscribe-auth/AB1cd6IC5DMCIW-FA6Wh3-RRjKpV3hRYks5rFC_sgaJpZM4LDrGF>
.
|
@aprokop has been issued an invitation to my fork.
…On Mon, Dec 5, 2016 at 11:09 AM, Adam J. Stewart ***@***.***> wrote:
I think it would be good to convert +fpic to +pic and convert -fPIC to
self.compiler.pic_flag in the same PR, as they touch a lot of the same
packages. @citibeth <https://github.com/citibeth> If it isn't set
already, can you allow @aprokop <https://github.com/aprokop> to push
directly to this PR?
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#2479 (comment)>, or mute
the thread
<https://github.com/notifications/unsubscribe-auth/AB1cd4CKWW1fAeQa6uCgYoeQxC0Ync-Hks5rFDcegaJpZM4LDrGF>
.
|
This is untested at this point.
Before my push, I did check for flake8, and fixed those except long lines. What's the policy on it? |
@aprokop Any commit will fail the CI tests if it does not satisfy flake8. |
@becker33 Then, is the flake8 feature was added just recently? 'Cause the failures are stemming from long lines already present in the repo. Edited: |
@aprokop I'm also confused as to why it is complaining about line 70 of superlu-dist, as that line isn't too long. The flake8 logic has been in place for a long time, and there should be no packages in Spack that fail. Btw, we do have some exceptions to the regular flake8 tests. They can be seen in |
@adamjstewart The way I tested locally was with ViM's nvie/vim-flake8 plugin. It did seem to highlight everything beyond 79 symbols in a line. For instance, it highlighted line 34 in scotch: url = "http://gforge.inria.fr/frs/download.php/latestfile/298/scotch_6.0.3.tar.gz" with an error
That line is already in the repo, so I am confused. Are there different versions of flake 8 (I'm not very familiar with it)? |
@aprokop We have flake8 configured to ignore any url lines, for example. @adamjstewart pointed you towards those changes correctly. |
@becker33 Thanks! I should start reading comments more carefully. I'll update the PR to fix the issues introduced by me. |
Here are the specific lines that ignore things like long URLs and long version() directives: If you want to test locally for flake8 compliance, simply use:
More info can be found in the documentation: |
@adamjstewart It's funny, but running
which is similar to the failing CI except for additional presence of yaml-cpp. |
That's very interesting, because |
Rebasing the branch on top of 3a15778f103e spack (test) $ spack flake8
=======================================================
flake8: running flake8 code checks on spack.
Modified files:
=======================================================
Flake8 checks were clean.
3a15778f103e spack (test) $ ./share/spack/qa/run-flake8-tests
Dependencies found.
=======================================================
flake8: running flake8 code checks on spack.
Modified files:
var/spack/repos/builtin/packages/adios/package.py
var/spack/repos/builtin/packages/cantera/package.py
var/spack/repos/builtin/packages/hdf/package.py
var/spack/repos/builtin/packages/metis/package.py
var/spack/repos/builtin/packages/mumps/package.py
var/spack/repos/builtin/packages/netcdf/package.py
var/spack/repos/builtin/packages/netlib-scalapack/package.py
var/spack/repos/builtin/packages/openblas/package.py
var/spack/repos/builtin/packages/otf2/package.py
var/spack/repos/builtin/packages/parallel-netcdf/package.py
var/spack/repos/builtin/packages/parmgridgen/package.py
var/spack/repos/builtin/packages/scorep/package.py
var/spack/repos/builtin/packages/scotch/package.py
var/spack/repos/builtin/packages/suite-sparse/package.py
var/spack/repos/builtin/packages/sundials/package.py
var/spack/repos/builtin/packages/superlu-dist/package.py
var/spack/repos/builtin/packages/superlu/package.py
var/spack/repos/builtin/packages/trilinos/package.py
var/spack/repos/builtin/packages/yaml-cpp/package.py
var/spack/repos/builtin/packages/zoltan/package.py
=======================================================
var/spack/repos/builtin/packages/scotch/package.py:98: [E501] line too long (81 > 79 characters)
var/spack/repos/builtin/packages/superlu-dist/package.py:68: [E501] line too long (103 > 79 characters)
var/spack/repos/builtin/packages/yaml-cpp/package.py:32: [E901] SyntaxError: unexpected character after line continuation character
Flake8 found errors. |
@adamjstewart Nevermind, my |
What else do we need for this PR? Should we just squash, rebase and push? |
Aside from rebasing, I don't see anything this PR is missing. I trust that you haven't missed any packages. Unfortunately, this PR has the side effect that none of these packages will be able to compile with the NAG compiler anymore. But I think I'm the only one who uses that anyway. |
@adamjstewart: how does this break NAG? |
Between the NAG issues and the next gen shared flag we have discussed, I am
not thrilled with this PR.
…On Dec 8, 2016 11:49 AM, "Todd Gamblin" ***@***.***> wrote:
@adamjstewart <https://github.com/adamjstewart>: how does this break NAG?
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#2479 (comment)>, or mute
the thread
<https://github.com/notifications/unsubscribe-auth/AB1cd6w1SxjenR8Wcd89jlku7B7Ocn0Cks5rGDUqgaJpZM4LDrGF>
.
|
@adamjstewart: two thoughts. One, we could provide separate |
Either would be fine by me. It's not an urgent need. The only thing I use NAG for is building HDF4/5 and NetCDF, but neither build with NAG at the moment, so I'm waiting for the next release to come out. |
@adamjstewart wrote:
There are few places around the place with hardcoded I guess the NAG issue is bothersome. How soon do people expect to properly solve the whole shared/static/pic flags issue? I wanted to participate in today's call but will have to skip it due to conflicts. Can somebody bring the issue up there, and maybe define steps to be done? |
Meh. I'll fix those as I encounter them. I don't build that many packages with NAG. |
@aprokop: it would be great to have you on the call. I am not sure we'll get the flags stuff done for |
@tgamblin Not sure yet. I have a recurring meeting at that time, but I'll see if I can skip it. |
Can I close this PR? I don't think it should be merged. I do think we should:
|
@citibeth So, just merge the patches I did that switch to |
Actually, most of the packages work correctly for me at the moment. The problem is that NAG (which only accepts Basically, this change will make things worse for NAG. But I think the intent is good. The end goal will be to be able to specify CC/CXX pic_flags and F77/FC pic_flags separately. I am not opposed to this change going in, as I'm probably the only one using NAG and it isn't able to build the HDF/NetCDF stack I want at the moment anyway. Although, I'm the one who requested the pic_flag, and I don't think |
@adamjstewart Out of curiosity, have you tried talking to NAG folks and see if they are open to adding support to |
I have not. @ThemosTsikas, are you aware of any plans for NAG to support |
So... does this mean that #2375 does not work with mixed toolchains? If so, I suggest we focus on fixing that issue, and then converting existing packages to use |
Hello, we have no plans to add -fPIC/-fpic options to the NAG Fortran Compiler. |
This is a decent start. Packages need to be checked individually to conform to #2375