FIX: grdinterpolate -E can't set correct unit

[Dengda98]

bug description

grdinterpolate -E can't set correct unit,

$ gmt grdinterpolate S362ANI_kmps.nc?vs -E0/60/110/60+i1d -T30/2890/10 -Gtest.nc  && gmt grd2xyz test.nc | head -n5
grdinterpolate [WARNING]: Your filename 0/60/110/60+i1d has what appears as valid GMT modifiers (from list +Udhinosu) but they are embedded rather than appended to the filename - modifiers ignored
0       2890    7.28660011292
112421.290721   2890    7.2931098938
224842.581442   2890    7.30029201508
337263.872163   2890    7.30822467804
449685.162883   2890    7.31660509109

(cube can be downloaded by curl -s http://ds.iris.edu/spudservice/data/17996723 -o S362ANI_kmps.nc)

Whats' changed

This part (-E) should be similar to that of grdtrack. In grdtrack, E.unit and projection scale are determined after the grid data reading, but this is opposite in grdinterpolate, see code below,

gmt/src/grdinterpolate.c

Lines 464 to 479 in 5beb2bd

	if (gmt_get_dist_units (GMT, Ctrl->E.lines, &Ctrl->E.unit, &Ctrl->E.mode)) { /* Bad mixing of units in -E specification */
	GMT_Report (API, GMT_MSG_ERROR, "Cannot mix units in -E (%s)\n", Ctrl->E.lines);
	Return (GMT_RUNTIME_ERROR);
	}
	if (gmt_init_distaz (GMT, Ctrl->E.unit, Ctrl->E.mode, GMT_MAP_DIST) == GMT_NOT_A_VALID_TYPE) /* Initialize the distance unit and scaling */
	Return (GMT_NOT_A_VALID_TYPE);
	/* Need to get dx,dy from one grid */
	if (Ctrl->Z.active) /* Get the first file */
	sprintf (file, "%s", Ctrl->In.file[0]);
	else /* Get the first layer from 3-D cube possibly via a selected variable */
	sprintf (file, "%s?%s[0]", Ctrl->In.file[0], cube_layer);
	if ((Grid = GMT_Read_Data (API, GMT_IS_GRID, GMT_IS_FILE, GMT_IS_SURFACE, GMT_CONTAINER_ONLY, NULL, file, NULL)) == NULL) {
	GMT_Report (API, GMT_MSG_ERROR, "Unable to read header from file %s.\n", file);
	Return (API->error);
	}

I use printf() to check the GMT->current.map.dist[GMT_MAP_DIST].scale before and after the GMT_Read_Data(), scale is redetermined. Grid here is just used to determined the default interval. So I just change the position of the calling function GMT_Read_Data() part, move upward a little bit.

The new result seems fine,

$ gmt grdinterpolate S362ANI_kmps.nc?vs -E0/60/110/60+i10k -T30/2890/10 -Gtest.nc  && gmt grd2xyz test.nc | head -n5
grdinterpolate [WARNING]: Your filename 0/60/110/60+i10k has what appears as valid GMT modifiers (from list +Udhinosu) but they are embedded rather than appended to the filename - modifiers ignored
0       2890    7.28660011292
10.030152332    2890    7.28715085983
20.0603046639   2890    7.2877073288
30.0904569959   2890    7.28826999664
40.1206093279   2890    7.28883838654

$ gmt grdinterpolate S362ANI_kmps.nc?vs -E0/60/110/60+i1.5d -T30/2890/10 -Gtest.nc  && gmt grd2xyz test.nc | head -n5
grdinterpolate [WARNING]: Your filename 0/60/110/60+i1.5d has what appears as valid GMT modifiers (from list +Udhinosu) but they are embedded rather than appended to the filename - modifiers ignored
0       2890    7.28660011292
1.51654172632   2890    7.29662418365
3.03308345265   2890    7.30822467804
4.54962517897   2890    7.3208527565
6.0661669053    2890    7.33400297165

$ gmt grdinterpolate S362ANI_kmps.nc?vs -E0/60/110/60+i30m -T30/2890/10 -Gtest.nc  && gmt grd2xyz test.nc | head -n5
grdinterpolate [WARNING]: Your filename 0/60/110/60+i30m has what appears as valid GMT modifiers (from list +Udhinosu) but they are embedded rather than appended to the filename - modifiers ignored
0       2890    7.28660011292
30.0181455108   2890    7.28972959518
60.0362910215   2890    7.29303884506
90.0544365323   2890    7.2965130806
120.072582043   2890    7.300137043

[joa-quim]

Looks good.