-
Notifications
You must be signed in to change notification settings - Fork 1
Conversation
Avoid confusion about primary repo when forking
Also, use MarkDown for hyperlink.
Merging pull request from Randall ( @codecurve ) to clarify the ReadMe to avoid possible confusion on the prime repository.
Just to be clear: review comments on cellml/libcellml#26 still apply to this pull request, but please continue any further discussion on this pull request. |
(Created as follows: "doxygen -g")
These are based on settings use on December's work, as well as ones that were on Hugh Sorby's pull request.
This is to show that the test actually does test something.
This reverts commit bd4a32c.
All lowercase, "_" separates words. See cellml/libcellml#30
Was in Model.h in previous code, lost since rollback.
Review progress Done (or otherwise addressed):
The following are not done yet. I'd propose we add all but the first one of these as a separate issue ("cmake improvements"), and defer it till later, so that we can focus on domain specific issues for now.
|
I don't think it is acceptable to delay addressing outstanding issues in the review. They all seem like either critical issues to address in the proposed code or trivial changes that should be quick to fix. |
Further review comment: I'd expect the documentation of the getVersion() method to include a description of the returned string. |
Clarification: in the src directory create a directory "libcellml" into which you put API header files |
Further to @hsorby's comment regarding "-g" in the CXX flags, it is bad to force debug build (and even worse to do so in a manner that will not work across all target compiler toolchains) - should make use of standard CMake build types (Debug, Release, etc) which will take care of such flags in a way that is safe across compiler toolchains. |
Test target only generated if `enable_testing()` is in top-level file.
Also for Build dir, and only where appropriate.
"check for cmake c++ compiler support is not well thought through, if there is a problem with the compiler it should be an error" is resolved if we use cmake 3.2 (currently at rc2 stage). See https://github.com/cellml/libcellml/blob/ab70fbb3798597fde0c23cad39c5df16df0cfefc/CMakeLists.txt |
I would probably use this form for setting a targets properties: set_target_properties(libcellmlTest PROPERTIES CXX_STANDARD 11 CXX_STANDARD_REQUIRED ON) I would prefer to stick with CMake 3.1 if possible. As demonstrated the above works on my GNU/Linux with CMake 3.1 |
BuildBot tests currently failing on Windows and Mac slaves, passing on Linux. The fix is to use CMake 3.2. Once CMake 3.2 is released (should be soon, since it is at release candidate 2 at the time of writing this), we will install it on the build slaves, and update the minimum version requirement to 3.2. Leaving minimum version requirement at 3.1 for now so that it still builds on Linux. See http://www.cmake.org/Bug/view.php?id=15355 This was discussed with @hsorby a few minutes ago. |
|
CodeSynthesis XSD is an XML bindings generator. This is the library to which generated bindings code will make calls. See http://www.codesynthesis.com/products/xsd/
@hsorby finished off the remaining cmake task (still to be checked?) and added Apache 2.0 license.
…n when CMake version 3.2 is available.
Commenting out setting of cxx standard to cxx-11, intending to add it back later.
…ult set by CMake.
Merge CodeSynthesis XSD library (take 2)
Changes required to get the tests passing on Buildbot
That
should not be here. |
With the mix up in merging #33 and reverting (#35) and its subsequent merge into this pull request, I wonder if it might not be significantly less hassle to simply start this one over? Is cherry picking 7ea7aab and starting from there an option? That seems to be the one with everything in a state which is working and presumably answering most issues that have arisen in this review... |
I haven't followed things very closely, but it certainly looks like a mess... :) |
Originally was pull request cellml/libcellml#26, but due to not being able to change the target branch of a pull request, had to close and restart (since original erroneously targeted master).
The focus here is on getting Google Test into the repo, so that it does not clutter future pull requests which focus on actual use cases.