-
Notifications
You must be signed in to change notification settings - Fork 0
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
Fix standalone build #1
Comments
@platipodium @josephzhang8 I could see |
BTW, i am not seeing any changes in my PR that removes |
@josephzhang8, @platipodium , @uturuncoglu I checked the latest version of SCHISM and it compiled successfuly only if I explicitly supply to cmake the option: -DBLD_STANDALONE=ON. This should be set ON by default. Maybe I missed it, why we need to have the BLD_STANDALONE flag? What is its purpose? |
@pvelissariou1 The build system was creating issue under UFS Coastal since it was not design to compile it under another application. So, I introduce BLD_STANDALONE=ON to get rid of those issues. The default is OFF but if we don't want to provide this option every time when we are trying to install SCHSIM, then I could change the default to ON and then make required change in the UFS Coastal SCHSIM build interface. |
@uturuncoglu Let's set BLD_STANDALONE=ON as the default and use BLD_STANDALONE=ON/OFF in UFS-Coastal depending upon the type of compilation. Still, I believe we need to address the NetCDF issues in a different way, without having this BLD_STANDALONE flag in place. What were the issues while building SCHISM in UFS-Coastal? I remember building SCHISM using previous versions of UFS-Coastal without any issues. Your thoughts? |
Thx @pvelissariou1 @uturuncoglu @platipodium . Adding standalone flag worked on my side. If it's hard to change the default, we can live with this by adding the flag inside SCHISM.local.build. |
@josephzhang8 My thinking is that, because most people use SCHISM in a standalone configuration it is better to have BLD_STANDALONE=ON by default. This is easily done within CMake. I haven't checked the GNU make configs yet. Do we need to implement the BLD_STANDALONE over there too? |
I am planning to get to this, by the way, but I've been hampered by a lot of hardware issues which seem to be finally clearing. Even if you do a workaround I'll try to take a look.
…________________________________
From: Joseph Zhang ***@***.***>
Sent: Friday, February 9, 2024 11:03 AM
To: oceanmodeling/schism ***@***.***>
Cc: Ateljevich, ***@***.*** ***@***.***>; Mention ***@***.***>
Subject: Re: [oceanmodeling/schism] Fix standalone build (Issue #1)
Thx @pvelissariou1<https://github.com/pvelissariou1> @uturuncoglu<https://github.com/uturuncoglu> @platipodium<https://github.com/platipodium> . Adding standalone flag worked on my side.
If it's hard to change the default, we can live with this by adding the flag inside SCHISM.local.build.
@water-e<https://github.com/water-e>
—
Reply to this email directly, view it on GitHub<#1 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AG2AJC6BGA6F6WL7DFF3DXTYSZXHZAVCNFSM6AAAAABC4O7LC6VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTSMZWGQ2TIMBUGU>.
You are receiving this because you were mentioned.Message ID: ***@***.***>
|
Thx @water-e |
I'll go ahead push the change to SCHISM repo, b/c many users are depending on cmake working. If we later on decide to set flag on UFS side we can revise. Thx |
Thx @pvelissariou1 for teh reminder on gcc. I just tested it on our HPC and it works also. My fix is to add STANDALONE inside cmake/SCHISM.local.build, together with other SCHISM options like OLDIO etc, except that it should not be turned off for SCHISM alone. |
@josephzhang8 @pvelissariou1 @water-e I think we could change the default in the SCHSIM build system. I think it will be easy to adjust the UFS Coastal. |
@uturuncoglu I completely agree. |
@uturuncoglu , @josephzhang8 Ufuk did you have the opportunity to compile with WWM ON? This configuration was failing during compilation. I'll check on this later today from my side. |
@pvelissariou1 No. Is this using internal WW3? We have a regression test working with external WW3 and it is working. |
@uturuncoglu ok, I will check and report back. WWM is the internal wave model. |
Will do.. thx @pvelissariou1 |
I'm able to compile with WWM on. |
Thank you Joseph. Let me try it within UFS-Coastal and I'll report back. |
@uturuncoglu , @josephzhang8 I compiled SCHISM standalone and in UFS-Coastal. The results are as follows (please read through):
NOTE: The SCHISM commit in UFS-Coastal is the same as in the standalone case (I replaced SCHISM in UFS-Coastal). The UFS executable contains SCHISM, CDEPS and CMEPS.
NOTE: The UFS executable contains SCHISM, WW3, CDEPS and CMEPS. My honest opinion is that the compilation approach of SCHISM (standalone or in UFS-Coastal) is not exactly what we want. We might need to improve on this. |
@uturuncoglu we can close, correct? |
@janahaddad Lets's keep it open for now. I need to double check. |
This commit fixes the issue related to non BLD_STANDALONE.
@josephzhang8 I have just tested recent changes in SCHSIM to fix the standalone build (schism-dev@eb562eb) and it is working under ufs-coastal side too. I run the RT and it passed. I'll sync out fork and update ufs-coastal then I'll close this ticket if this is also fine for you. |
Thx Kijin, @uturuncoglu for your help! I've also tested on our system and the new changes in cmake are fine. |
@josephzhang8 Okay. Great. I synced the SCHSIM under ufs-costal. @kjnam Thanks again for your help. |
@uturuncoglu @josephzhang8 Happy to be of help. |
@josephzhang8 @kjnam I am opening this issue since
and also model crashes with following error,
I think that second error is due to the first one. I am not sure why Anyway, here are the details of configurations,
Anyway, the only extra flag for SCHSIM is |
@janahaddad I reopened this ticket since last sync is breaking the ww3 coupling. Maybe cap needs to be adjusted but not sure at this point. Still investigating. I will point the working hash (f9b42db) in ufs-coastal until we solve the issue. |
@josephzhang8 I think there is a incompatibility issue between SCHSIM and SCHSIM-ESMF repository. Here is the line that prints out first WARNING message, https://github.com/schism-dev/schism-esmf/blob/9fc62b21afc1edcb691bef5ffa5b309375d5b07c/src/schism/schism_esmf_util.F90#L1127. It indicates that |
@josephzhang8 @platipodium Okay. I seems that the issue was arising due to recent changes that brings wave related fields to the output. I created a new issue for it #4. @platipodium I'll also create a PR to the NUOPC cap that fixes the |
@josephzhang8 @platipodium I created a PR in upstream that fixes the couple of issues. schism-dev/schism-esmf#29. I think I could close this issue but prefer to wait until PR is merged. |
Merged upstream. THank you for the hotfix |
@uturuncoglu |
@uturuncoglu The changes in param.nml are documented live inside src/Readme.beta_notes. In particular pay attention to 'remove' b/c of the way .nml works (you can omit parameters but cannot include those that are not in the list). |
@josephzhang8 @platipodium As I know this issue is fixed. If you don't mind could you confirm. So, I could close the issue. |
I think so. Thx @uturuncoglu ! |
The changes that are done to build model under UFS Coastal are creating issue in standalone build outside of the UFS Coastal. @josephzhang8 test threw a lot of errors related to netcdf and util scripts. There is a need for a target that does not compile any util scripts. Need also check
pschism
target.The text was updated successfully, but these errors were encountered: