<div dir="ltr"><div>Hi all,</div><div><br></div><div>The commit has been made. The masters of these VLAB Gerrit git repositories have been updated: NEMS, NCEPLIBS_pyprodutil, NEMSfv3gfs, EMC_NEMSGSM, EMC_ATM-WW3, EMC_HYCOM-GSM-CICE, and EMC_FV3-MOM6-CICE5. <br></div><div><br></div><div>I'm still debugging the failed compset to see if I can reproduce the
issue. All attempts to re-run the compset have succeeded thus far.</div><div><br></div><div>Sincerely,</div><div>Sam Trahan<br></div></div><br><div class="gmail_quote"><div dir="ltr">On Mon, 29 Oct 2018 at 13:27, Samuel Trahan - NOAA Affiliate <<a href="mailto:samuel.trahan@noaa.gov">samuel.trahan@noaa.gov</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr"><div dir="ltr"><div>Hi all,</div><div><br></div><div>The application developer told me to go ahead with this commit, so I'll commit it between 5-6 PM Eastern time today. I'm going to keep debugging, but I must reproduce the problem before I can determine its cause. Three different accounts were able to rerun the test with no problem, so the problem may be fugacious, or worse, sporadic.<br></div><div><br></div><div>Lastly, I have two updates to my original email about the commit:<br></div><div><br></div><div>1. I mentioned "odin support" was added in the commit. It is not. Theia support for non-Intel compilers is added in the commit.</div><div><br></div><div>2. The NCEPLIBS_pyprodutil submodule is updated. This adds a ".gitignore" that will ignore the Python *.pyc files. The NEMS level already had this, but that ".gitignore" lost its potency when the produtil was moved to a submodule. (A ".gitignore" is not applied recursively because each submodule is its own git repository.) Thanks to Dusan Jovic for noticing this problem.<br></div><div><br></div><div>Sincerely,</div><div>Sam Trahan<br></div></div></div><br><div class="gmail_quote"><div dir="ltr">On Mon, 29 Oct 2018 at 11:36, Samuel Trahan - NOAA Affiliate <<a href="mailto:samuel.trahan@noaa.gov" target="_blank">samuel.trahan@noaa.gov</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr"><div>Hi all,</div><div><br></div><div>In order to debug this problem, I need to run a number of tests on Theia. All projects accessible to emc.nemspara or myself have hit their monthly CPU allocation. That means I cannot even start debugging this until Thursday, when the allocations reset. Most likely, this will delay the commit until next week. I will let you know if I find another solution.<br></div><div><br></div><div>Sincerely,</div><div>Sam Trahan<br></div></div><br><div class="gmail_quote"><div dir="ltr">On Mon, 29 Oct 2018 at 10:57, Samuel Trahan - NOAA Affiliate <<a href="mailto:samuel.trahan@noaa.gov" target="_blank">samuel.trahan@noaa.gov</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr"><div dir="ltr"><div>Hi all,</div><div><br></div><div>I have finished running the regression test system for this commit. One of the compsets for FV3-MOM6-CICE5 failed on Theia, and all other compsets for all apps succeeded everywhere else. I'm having the app code manager look into the failed case. I don't think it is related to these changes, but I want the app's code manager to decide that.</div><div><br></div><div>Test results are here:</div><div><br></div><div><a href="http://www.emc.ncep.noaa.gov/projects/nems-commit/build-update-wave2/" target="_blank">http://www.emc.ncep.noaa.gov/projects/nems-commit/build-update-wave2/</a></div><div><br></div><div>The tests report no failure because I reran the failed compset, and it succeeded.<br></div><div><br></div><div>Sincerely,</div><div>Sam Trahan<br></div></div></div><br><div class="gmail_quote"><div dir="ltr">On Fri, 26 Oct 2018 at 12:20, Samuel Trahan - NOAA Affiliate <<a href="mailto:samuel.trahan@noaa.gov" target="_blank">samuel.trahan@noaa.gov</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr"><div dir="ltr"><div dir="ltr"><div dir="ltr"><div dir="ltr"><div dir="ltr"><div dir="ltr"><div dir="ltr"><div dir="ltr"><div dir="ltr"><div>Hi all,</div><div><br></div><div>Sometime Monday evening, I will commit the second wave of build system changes: bug fixes and feature additions to the new NEMS build system. Final tests are beginning now. With these changes, DATM and GSDCHEM are confirmed to work, and several reported bugs are fixed. We're half-way to getting CCPP fully working; some changes are still needed in FV3, and those are waiting for other commits.<br></div><div><br></div><div>Most of the changes in this batch of commits are from Dom Heinzeller, who is adding support for compiling on non-NOAA machines, and for running with CCPP. I'm only committing the NEMS and NEMSfv3gfs parts of his changes; he has some changes in FMS and FV3 which will be committed later, hopefully late next week.<br></div><div><br></div><div><br></div><div></div><div>The branches in NEMS and NEMSfv3gfs are build-update-wave2. For other apps, the master is used.<br></div><div><br></div><div><br></div><div>Changes that are NOT from Dom Heinzeller:<br></div><div><br></div><div>1. Remove an extra "clean" step in NEMS/src/incmake/component_FV3.mk. This fixes a bug reported by Ratko and Dusan wherein FV3 would always be cleaned, even if you asked for it not to be.<br></div><div><br></div><div>2. Do not delete the *.tmp.f90 files when building NEMS. This is why NEMS always rebuilt its source code. Dusan provided this fix.</div><div><br></div><div>3. GSDCHEM clean rule: do not fail to clean if the Makefile is absent. This fix is from Raffaele Montuoro.</div><div><br></div><div>4. Use -j1 when building Wavewatch 3. This is a temporary workaround for a Wavewatch3 build system issue. A better solution is in progress.<br></div><div><br></div><div>5. In the build system documentation, add a big box reminding you to "cd NEMS" before building.</div><div><br></div><div><br></div><div>Dom Heinzeller's changes:</div><div><br></div><div>1. Rename "macos" build targets to "macosx" for consistency with components.<br></div><div><br></div><div>2. Set appropriate default NEMS_COMPILER values for various platforms.<br></div><div><br></div><div>3. Skip the module-setup.sh.inc if the platform does not support the "module" command.</div><div><br></div><div>4. Add ccpp/include to the include flags in the CCPP's ESMF makefile fragment. (The NEMS build system treats CCPP as a component.)<br></div><div><br></div><div>5. Update linux, macosx, odin, and cheyenne build environment logic so it actually works. (FV3 and FMS commits are still needed for full support.)</div><div><br></div><div>6. Update PGI & GNU compiler suite logic so it actually works. (FV3 and FMS commits are still needed for full support.)<br></div><div><br></div><div>7. ccpp/physics/physics/sfcsub.F update of code to compile with gfortran v8+</div><div><br></div><div>8. Documentation updates for porting and CCPP.</div><div><br></div><div><br></div><div>Sincerely,</div><div>Sam Trahan<br></div></div></div></div></div></div></div></div></div></div></div>
</blockquote></div>
</blockquote></div>
</blockquote></div>
</blockquote></div>