-
Notifications
You must be signed in to change notification settings - Fork 358
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
Refactor Connection Generator Interface #483
Conversation
…to avoid confusion. - "Phase 6" heading now in one place, some info given about which test scheme is used. - Check for nosetest commented.
Improve TravisCI output by displaying files with warnings.
This library contains all the code necessary for GNUReadline interaction. We linke it only to sli and nest executable, so that for example pynest is not affected by readline linking stuff.
and add implicit dependency to nestutil explicitly
…module.cpp and conngen.{h,cpp} directly
…ct_gidcollection+refactor_cmake
- Remove the variants of CGConnect that take subnets - Refactor the variants of CGConnect to take gidcollections and (via conversions in SLI) also arrays and intvectors - Remove convenience wrappers for internal function cgnext - Adapt PyNEST wrapper for CGConnect - Adapt and extend the test for CGConnect
Please note that this test is not yet functional due to missing functionality in libcsa
…factor_conngen_interface
@heplesser: Thanks for the thorough review. I've addressed your minor comments in the latest commits and will look at the more complex ones either later today or tomorrow. This is what's missing:
Regarding your issues with libneurosim: I've seen the same problem on JURECA last week and will investigate further as soon as the machine is back from maintenance. I'll report back when I know more. |
Just an additional note: this should probably not be merged before #817 was addressed. Currently, we are likely to miss proper testing of libneurosim/CSA related code. |
@heplesser: I have successfully reproduced your problem with libneurosim. Contrary to what I said before, I now think that this is a proper bug and not something that only occurs on some machines. I've opened #821 with some more explanations and a question for you. |
@heplesser I have addressed your remaining issues in the latest commits.
9601223 adds a Doxygen comment for
70d77c4 removes the copying.
The
This answer on Stack Overflow says that this guarantee was missing in C++98 but added later. I assume it is safe enough as I can only think of rather complicated ways a compiler should implement a vector non-contiguously given all its other properties.
156b88b adds a check and a corresponding exception for this.
a835bd4 converts the test to an example. |
Please note that my latest commits are rather untested due to INCF/csa#6. |
@jougs Thanks for your fixes! Travis was unstable, I have restarted the builds there (failed again). Concerning the doxygen comment placement, we can leave it where it is now, but generally my understanding is that class and method documentation should be in the header file, since the header might be all that developers see (a bit of closed source thinking here, but I think it makes sense). Concerning the testing: How much work would it be to create a very simple connection generation code, so that one at least would have basic testing in place? |
@jougs Thanks to advice from @lekshmideepu I now realised that the problem causing Travis to fail is that the See also CSA Issue 7. |
@jougs With CSA issue 7 fixed it looks pretty fine now, just some clang-formatting issues to fix. |
Thanks for the review. I've addressed the remaining issues as follows: |
cdec295
to
5898e37
Compare
@jougs Great work! But I am confused after a look at the test configuration including libneurosim: There is says (l 5868ff):
So the tests in |
Well done! And merging ... |
Many thanks to all who made this long story come to a good end. What a relief :-) |
In particular this PR
cg_connect.{h,cpp}
files to theconngen.{h,cpp}
filesCGConnect
variants to supportintvectordatum
andgidcollection
istead ofsubnet
I propose @mdjurfeldt and @uahic as reviewers.