[R-sig-Geo] Error in CRS("EPSG:4326") : no arguments in initialization list

[Tomislav Hengl]

OK got it.

I've resubmitted to CRAN so hopefully this time it will work on all 
versions.


On 4/29/21 3:13 PM, Roger Bivand wrote:
> On Thu, 29 Apr 2021, Tomislav Hengl wrote:
> 
>>
>> Thanks Roger. If I understand correctly, I should simply replace 
>> 'CRS("EPSG:4326")' with 'CRS("+proj=longlat +datum=WGS84 +no_defs")'. 
>> Otherwise the EPSG codes should work in R with any version correct?
> 
> No, just assign "+proj=longlat +datum=WGS84 +no_defs", no CRS(), no 
> CRSargs() at all, so preventing CRS() reaching out through rgdal to 
> PROJ. It probably shouldn't be canned anyway, but should rather be 
> generated at run time not install time (and CRSargs() should be 
> deprecated as it removes WKT2 representations). Note that +datum=WGS84 
> is very imprecise, as it varies by epoch, and is from EPSG 10 a datum 
> ensemble rather than a datum.
> 
> EPSG codes given as CRS("EPSG:4326") work sp >= 1.4 and rgdal >= 1.5 
> with PROJ >= 6 and GDAL >= 3. The puzzle here is how CRS() mistakes the 
> setting and tries to retrieve "EPSG:4326" from PROJ < 6.
> 
> Roger
> 
>>
>>
>> On 4/28/21 8:48 PM, Roger Bivand wrote:
>>>  Tom,
>>>
>>>  Your package code itself runs lots of stuff in R/AAA.R that it doesn't
>>>  need to. Here you could simply put the output string of:
>>>
>>>>  rgdal::CRSargs(sp::CRS("EPSG:4326"))
>>>  [1] "+proj=longlat +datum=WGS84 +no_defs"
>>>
>>>  instead of going out to sp/rgdal/PROJ. In this case, sp and rgdal
>>>  shouldn't let you do this on a system running PROJ 5.2.0 (and I'll look
>>>  at that), but your work-around is the bare Proj4 string, which is the
>>>  same in any case. The error message is from PROJ, returned from rgdal
>>>  to sp. If you want to condition on PROJ version, you can do that, but
>>>  instantiating CRS is costly, and here lazy-loading is trying to save
>>>  time by pre-filling the value.
>>>
>>>  If this isn't enough, please let me know and I'll install the same PROJ
>>>  and GDAL versions as used for Solaris checks locally, and try to
>>>  resolve the problem in those packages. That will however take time. I'm
>>>  assuming that sp and rgdal are the current CRAN versions 1.4-5 and
>>>  1.5-23 on Solaris, both of those pass on that platform.
>>>
>>>  Hope this helps,
>>>
>>>  Roger
>>>
>>>  On Wed, 28 Apr 2021, Tomislav Hengl wrote:
>>>
>>>>
>>>>  Hi Roger et al,
>>>>
>>>>  We are experiencing some problems with committing plotKML to CRAN.
>>>>  Automated check reports:
>>>>
>>>>  "Error in CRS("EPSG:4326") : no arguments in initialization list"
>>>>
>>>>  See:
>>>>
>>>>  https://www.r-project.org/nosvn/R.check/r-patched-solaris-x86/plotKML-00install.html 
>>>>
>>>>  https://github.com/Envirometrix/plotKML/issues/5
>>>>
>>>>  I assume it has to do with the proj / rgdal versions? Should I switch
>>>>  to using again CRS("+init=epsg:4326")? Seems that different operating
>>>>  systems now have different issues.
>>>>
>>>>  On R-forge plotKML installs without problems:
>>>>
>>>>  https://r-forge.r-project.org/R/?group_id=1106
>>>>
>>>>  _______________________________________________
>>>>  R-sig-Geo mailing list
>>>>  R-sig-Geo using r-project.org
>>>>  https://stat.ethz.ch/mailman/listinfo/r-sig-geo
>>>>
>>>
>>
>