lambert proj appears to be broken??

[doutriaux1]

old lambert over north america:

p=x.createprojection()
p.type = "lambert"
b=x.createboxfill()
b.projection=p
x.plot(s(longitude=(-120,-20),latitude=(20,60)),b)

[doutriaux1]

new vcs

[doutriaux1]

@aashish24 @danlipsa if you beat me to it please feel free to fix it.

[doutriaux1]

ok making progress it looks like gctp was smarter than proj4 we need to set all parameters manually or it will not work

p.centralmeridian = -80
p.originlatitude=0
p.standardparallel1 = 20
p.standardparallel2=60

[doutriaux1]

fixed in 2.4.0 but on macs exposes proj4 issue listed @ #1777

[danlipsa]

@doutriaux1 @aashish24 Do we need to pass +over to the lambert projection (or to polar (non gctp))? What I want to get at, is a test case, where we need to pass +over, it works on most machines but it does not work on certain macs and annie. If we don't have that we'll spend time to fix something that we don't really need. Maybe for these projections, +over does not work consistently because people don't need that so nobody tested for that.

[doutriaux1]

it's worth trying. I will do that now.

[doutriaux1]

@danlipsa I don't think that's the issue, looking at the code:

no_over_proj4_parameter_projections = round_projections+["aeqd", "lambert conformal c"]

Then we have a section in vcs2vtk.py that does:

        if projection.type not in no_over_proj4_parameter_projections:
            pd.SetOptionalParameter("over", "true")
        else:
            pd.SetOptionalParameter("over", "false")

So obviously we're not trying over on lambert and it still fails on macs.

[doutriaux1]

i think the issue could be coming from:

            setProjectionParameters(pd, projection)

because basically it seems like that it is the only thing we took out for poalr non gctp that made it work again, but going into the function I don't think it does anything...

[danlipsa]

@doutriaux1 What is the test that is failing on macs? Is it failing on annie as well?

[doutriaux1]

only on mac, if you look at CMakeList.txt it is commented out if it's an apple. test_vcs_lambert I believe is the name. I "think" it passes on annie, but it could also be luck 😉

[danlipsa]

Interesting. I tried this test on my mac - which had problems with +over and polar non gctp and it passes. I will create a branch where I activate this test and see where it fails.