View Issue Details
| ID | Project | Category | View Status | Date Submitted | Last Update |
|---|---|---|---|---|---|
| 0002767 | OpenFOAM | Contribution | public | 2017-11-21 14:09 | 2017-11-21 22:00 |
| Reporter | Pfiffikus | Assigned To | henry | ||
| Priority | low | Severity | feature | Reproducibility | always |
| Status | closed | Resolution | no change required | ||
| Summary | 0002767: prevent contradicting compiler directives for WM_PRECISION_OPTION: wm_sp, wm_dp, or wm_ldp | ||||
| Description | What about to detect contradicting compiler directives by #if (defined WM_SP && (defined WM_DP || defined WM_LDP)) || (defined WM_DP && (defined WM_SP || defined WM_LDP)) || (defined WM_LDP && (defined WM_SP || defined WM_DP)) #error You cannot use more than only one of WM_SP, WM_DP, WM_LDP at the same time! #endif #if !(defined WM_SP || defined WM_DP || defined WM_LDP) #error You must use either WM_SP, WM_DP, or WM_LDP #endif | ||||
| Steps To Reproduce | Try to link your code, compiled with -DWM_SP, with initially installed sources, compiled with -DWM_LP. It could fail due to unresolved references ... | ||||
| Additional Information | Yes, the prevention is only for safety in case of manual modification of make mechanism. | ||||
| Tags | compilation, linking | ||||
|
|
It is not clear where in the code you want this test or why. WM_LDP does not exist in OpenFOAM, did you add it? What does it do? |
|
|
@Henry: After a quick Google search, it looks like "WM_LDP" was implemented in the fork "foam-extend", which refers to "long double precision". I have no idea what good would it do to have that much precision, but I guess it would be for planetary-sized simulations or for atomic/molecular modelling? As to why people would want to define these options manually? It's a trend that has been showing up on the forums once in a while, people compiling and linking other code directly to OpenFOAM libraries, without relying on OpenFOAM's wmake stack. @Pfiffikus: The use of those macro checks does not check whether the code was first compiled with another option; for example, if a cluster administrator installs OpenFOAM in a way commonly used on Linux distributions - e.g. having the libraries in "/usr/lib/OpenFOAM", instead of something like "/opt/OpenFOAM/5/platforms/linux64GccDPOpt" - then having these macro checks will not verify which options were originally used for compiling the libraries. ---- On the other hand, checking whether at least "WM_SP" or "WM_DP" is defined, could be useful for those trying to implement their own Makefile implementation... or if someone messes up the settings in the "rules" folder. I guess it would be enough to define this in the main 'Scalar.H' file, for it to always trigger the check when necessary. |
|
|
I don't see the need for this change and the proposal is not compatible with OpenFOAM anyway. I think it would be better to set and check these variable is the build scripts rather than adding this kind of clutter to the code. |
| Date Modified | Username | Field | Change |
|---|---|---|---|
| 2017-11-21 14:09 | Pfiffikus | New Issue | |
| 2017-11-21 14:09 | Pfiffikus | Tag Attached: compilation | |
| 2017-11-21 14:09 | Pfiffikus | Tag Attached: linking | |
| 2017-11-21 14:33 | henry | Note Added: 0009056 | |
| 2017-11-21 21:52 | wyldckat | Note Added: 0009057 | |
| 2017-11-21 22:00 | henry | Assigned To | => henry |
| 2017-11-21 22:00 | henry | Status | new => closed |
| 2017-11-21 22:00 | henry | Resolution | open => no change required |
| 2017-11-21 22:00 | henry | Note Added: 0009058 |
