[SHOT] update to latest master commit #3290
Conversation
|
@andreaslundell, it wasn't that easy unfortunately... |
|
|
|
Yeah, I'll push a commit that disables it. |
| - ENDIF( ${variable} ) | ||
| + # CHECK_CXX_SOURCE_RUNS("${source}" ${variable} ) | ||
| + # IF( ${variable} ) | ||
| + SET(${variable} 1) |
There was a problem hiding this comment.
So this just assumes that the answer to "did the code run" is "yes."
It might be safer to make this a "no"?
It's for things like these variables:
https://github.com/coin-or/CppAD/blob/e6e49284de7da18f936ebcbca0396150bbb1c48d/include/cppad/core/check_for_nan.hpp#L137-L146
There was a problem hiding this comment.
The answer may be platform-dependent, you have to see case-by-case. If the point of CPPAD_HAS_MKSTEMP is to check whether mkstemp is available, that should be a compiler check, no need to run anything.
|
@andreaslundell, I guess Windows doesn't like this: |
|
That seemed to fix the Windows issue. I don't know the root cause. Upstreamed to coin-or/SHOT#130 |
|
That's weird, but renaming enums should be harmless enough |
* [SHOT] update to latest master commit * Potential fix for run_source_test * Take 2 * Take 3 * Take 4
The commit we build as v1.0.1 has some bugs (jump-dev/AmplNLWriter.jl#142) that may have been fixed in the 283 commits to master since then :)
Since the current v1.0.1 JLL release isn't really usable from JuMP, we may as well try a v1.0.1+1 release with the latest commit that also includes the upstreamed patches.