Update vtk proj4 9 2 #1827
Update vtk proj4 9 2 #1827
Conversation
danlipsa
force-pushed
the
update_vtk_proj4_9_2
branch
3 times, most recently
from
0e0d84c
to
7bae2ac
Compare
|
@doutriaux1 Is cont-int/LLNL/Linux-RH6 (FULL) working? Because it says: 'checking whether the C compiler works... no' I would think it is a problem with the machine. Are there any mac buildbots left? |
|
@aashish24 @doutriaux1 Please review. I think this is a step in the right direction with aeqd and no other projection seems worst than it was. Some of the aeqd tests look a lot better some don't. Given that aeqd was completely broken in the past I would say to merge this and do another pass through the aeqd tests in the future. I think we can improve how they look. |
|
@aashish24 We need to merge proj4 upgrade branch into VTK first before we can merge this. See |
|
@danlipsa looks like xfvb needs a restart on RH6 (FULL) and bots are down after one build as usual... |
|
@danlipsa are you using Kitware/VTK? I thought we were supposed to use UV-CDAT/VTK so we do not have to wait for Kitware/VTK. |
Reviewed it and +2 it. @doutriaux1 once this goes into master, we will merge master into UVCDAT/VTK:uvcdat-master. |
|
@aashish24 I get conflicts merging VTK master into uvcdat-master. I'll let you do this merge as you have managed the commits into both. @aashish24 @doutriaux1 In the future I would suggest always merge things into VTK master before we use a certain SHA into uvcdat. Having a separate branch and merging things into both VTK master and uvcdat-master can lead to conflicts and makes it more difficult to update VTK. Note that ParaView used a separate branch in the past but it got rid of it and now it uses directly a certain SHA from VTK. |
|
@danlipsa things got really bad on my ubunut see: |
|
@doutriaux1 this looks like the old proj4. Make sure the VTK got updated: remove VTK-prefix and build/VTK-build |
You got merge conflict because I pull in one fix from UV-CDAT to VTK. I will fix the conflict as its a easy one. We pretty much have all the changes in VTK master except the FFPEG one. If we keep merging changes to VTK propoer, we won't have these conflicts. |
|
I'm pretty sure it was a fresh build, but I will double check |
|
@aashish24 OK, so after we merge everything into VTK we can just update uvcdat-master to the latest VTK isn't it? |
|
@doutriaux1 I will try running the tests on my mac as well. |
Right that is the idea. The only part thats not in VTK proper is the FFMPEG code. The reason I am afraid to bring it in VTK is because I am not 100% sure why the code was written the way it was written but somehow was working. I guess I will take the risk and push a VTK PR and see what happens. In the mean time, what I am going to do is reset UVCDAT/uvcdat-master to VTK master and bring in my change on top of it. |
danlipsa
force-pushed
the
update_vtk_proj4_9_2
branch
from
7bae2ac
to
c1bb819
Compare
|
@doutriaux1 @aashish24 All tests pass on my ubuntu, also cdat-web works. I'll run the tests at home on my mac. |
|
@danlipsa running this one more time on my mac just to be sure. |
|
@doutriaux1 I run this on my mac. Looks good besides expected (and anoying) small differences in some aeqd new images. vcs_mintics has more differences but think this is a test @chaosphere2112 just added. I added the file on linux as it was missing. |
|
@aashish24 I can remove the gdal stuff. Do need that saved in a patch file? |
Yes, thanks. A patch would work (just in case). |
danlipsa
force-pushed
the
update_vtk_proj4_9_2
branch
from
c1bb819
to
d690a69
Compare
…revious size. This causes test_vcs_configurator_resize to fail.
danlipsa
force-pushed
the
update_vtk_proj4_9_2
branch
from
d690a69
to
daf552b
Compare
|
@aashish24 Here is the patch: |
|
@doutriaux1 @aashish24 This is ready to merge! |
|
LGTM 👍 |
This is using a VTK branch not commited into master so do not merge yet.