Conversation
|
@kostrzewa and @chaoos :
|
cc527c1 to
c86f934
Compare
2d50b54 to
8f39dd2
Compare
There was a problem hiding this comment.
Even with the changes from my suggestions below, the build still fails for me with various compiler errors and warnings, like e.g.:
/home/roman/projects/tmp/tmLQCD/src/lib/io/gauge_read_binary.c: In function ‘read_binary_gauge_data’:
/home/roman/projects/tmp/tmLQCD/src/lib/io/gauge_read_binary.c:72:17: error: ‘g_cart_grid’ undeclared (first use in this function); did you mean ‘g_cart_id’?
72 | MPI_Barrier(g_cart_grid);
| ^~~~~~~~~~~
| g_cart_id
/home/roman/projects/tmp/tmLQCD/src/lib/io/gauge_read_binary.c:72:17: note: each undeclared identifier is reported only once for each function it appears in
/home/roman/projects/tmp/tmLQCD/src/lib/io/utils_construct_writer.c: In function ‘construct_writer’:
/home/roman/projects/tmp/tmLQCD/src/lib/io/utils_construct_writer.c:17:28: error: ‘g_cart_grid’ undeclared (first use in this function); did you mean ‘g_cart_id’?
17 | status = MPI_File_open(g_cart_grid, filename,
| ^~~~~~~~~~~
| g_cart_id
One can put these variables as extern, but then at link time it fail to resolve with undefined reference to g_cart_grid'`.
Apologies, if I build lemon with the new cmake system as static library, I can build tmlqcd properly. |
|
Sorry, the current version does not support the lemon configure build as the lemon cmake build system makes this more reliable. I need to check if I generate a pkg-config file for lemon |
| endif() | ||
|
|
||
| # check for fftw3 (rely on pkgconfig). | ||
| if(TM_USE_FFTW) |
|
remove LAPH_ev.c |
| set_target_properties(hmc PROPERTIES VERSION ${PROJECT_VERSION} SOVERSION 1) | ||
|
|
||
| # define a library and add the dependencies | ||
| target_link_libraries( |
|
please add a couple of example cmake calls to a |
|
I'm afraid some compilation paths have been broken. For example,
does not compile because the |
|
This happened during e591af5 and I think See |
|
We could add an additional github workflow which builds (but does not run) a set of compilation flag combinations. |
|
|
- Restored hopping.h - Rename hmc with tmlqcd - Fix TM_USE_DDalphaAMG when necessary - Remove header guards in git_hash.c
f109115 to
07f9d45
Compare
Co-authored-by: chaoos <chaoos@users.noreply.github.com>
- Restored hopping.h - Rename hmc with tmlqcd - Fix TM_USE_DDalphaAMG when necessary - Remove header guards in git_hash.c - Update Spack package
…ge.py Co-authored-by: chaoos <chaoos@users.noreply.github.com>
|
@kostrzewa: I removed the duplicated file |
|
It would be great if we could try to avoid force-pushes if possible. For example the work in #666 has been based on |
|
Note that in Regarding your comment about the insertion of
|
can we reach this with inlining. I may not want to include this in this PR though. I modified the last remaning file as well |
f27f6ba to
1a94df9
Compare
let's not include it here
okay, should I close the corresponding PR then? #672 |
|
I'm going to continue testing the build on different machines. It will take a few days but I would like to complete this before easter as the teaching period starts again on the 13th of April and I'll have too much to do then... |
[ignore] Actually it might be better to merge your PR first and then I can rebase afterwards. This fix needs to be applied to the last version as well. I merged your PR and I will fix potential conflicts. I need to check that I did not delete the hopping.h file in the "old" tree. |
include hopping.h and hoping_body_dbl.inc in tm_[sub,times]_Hopping_Matrix.c
|
A short recap of changes
|
|
i opened a PR in the lemon github repo updating the version to 1.2.0. I can either merge it manually and then create the tag accordingly or let someone else do it |
no it's not, we should probably also restore it there and then tag that as rel-6-0-2 the cmake_support branch will then become rel-6-1-0 |
I am already working on it. I will open a PR in a short while. |
|
Pr opened to fix |
tmlQCD can be compiled with CMake. So far the build supports