make check fails

[strk]

I suspect it has to do with setting PROJ_LIB in Makefile, I'll give that a try.

============================================
Running ./testIGNF using ../src/cs2cs:
============================================
doing tests into file proj_outIGNF, please wait
Rel. 4.7.1, 23 September 2009
<lt-cs2cs>: while processing file: <stdin>, line 1
pj_transform(): failed to load datum shift file
....

[strk]

setting PROJ_DEBUG and PROJ_LIB manually shows that the IGFN file is loaded successfully. But still the obtained transformations do not match the expected ones:

diff proj_outIGNF with proj_outIGNF.dist
2c2
<  3.300866856 43.4477976569 0.0000     *       * 0.000
---
>  3.300866856 43.4477976569 0.0000     3d18'0.915"E    43d26'52.077"N 0.000
4,11c4,11
<  600000.0000 2600545.4523  0.0000     *       * 0.000
<  135638.3592 2418760.4094  0.0000     *       * 0.000
<  998137.3947 2413822.2844  0.0000     *       * 0.000
<  600000.0000 2200000.0000  0.0000     *       * 0.000
<  311552.5340 1906457.4840  0.0000     *       * 0.000
<  960488.4138 1910172.8812  0.0000     *       * 0.000
<  600000.0000 1699510.8340  0.0000     *       * 0.000
< 1203792.5981 626873.17210  0.0000     *       * 0.000
---
>  600000.0000 2600545.4523  0.0000     652760.737      7033791.244 0.000
>  135638.3592 2418760.4094  0.0000     187194.062      6855928.882 0.000
>  998137.3947 2413822.2844  0.0000     1049052.258     6843776.562 0.000
>  600000.0000 2200000.0000  0.0000     649398.872      6633524.192 0.000
>  311552.5340 1906457.4840  0.0000     358799.172      6342652.486 0.000
>  960488.4138 1910172.8812  0.0000     1007068.686     6340907.237 0.000

The "null" gridshift fails loading (should it be attempted to be loaded at all?):

pj_open_lib(./IGNF): call fopen(./IGNF) - succeeded
pj_open_lib(null): call fopen(/usr/src/proj/proj/proj/nad/null) - failed

[warmerdam]

I observe that this failure started after r2122. Digging...

[warmerdam]

Didier,

I'm confused by some items in r2122. For instance, why do you include both +nadgrids=@null and +towgs84=0,0,0 in some coordinate systems? This is overspecified and might cause problems particular if in the future an interpretation is given to having both.

I think the failure of the tests is due to you changing the transformations slightly (nadgrids=@null is not the same as +towgs84=0,0,0 as perhaps you realized). I'd appreciate it if you could check the new testIGNF output results and commit them if applicable.