[R-pkg-devel] GitHub Action failed re-building 'sos.Rnw'

Spencer Graves @pencer@gr@ve@ @end|ng |rom e||ect|vede|en@e@org
Tue May 25 23:06:25 CEST 2021


Hi, Duncan:


	  I apologize for bothering you with such an obvious thing as using 
"CRAN() before library(sos)".


	  Now I have another problem:  "zq <- ???spline(2)" now retrieves the 
default 100 pages, not the 2 requested.


	  I think the processing of "???" has changed between 4.0.3 and 4.1.0. 
  I've done "debug(`?`)" in both 4.0.3 and 4.1.0:  The condition 
(!is.call(topicExpr1) || length(topicExpr1) != 2 || topicExpr1[[1]] !=
     "?" || !is.call(topicExpr1[[2]]) || length(topicExpr1[[2]]) !=
     2 || topicExpr1[[2]][[1]] != "?") evaluates initially to FALSE in 
4.0.3 but TRUE in 4.1.0.


	  You graciously contributed that code several years ago when I changed 
the name of the package from RSiteSearch to sos, and it has worked 
wonderfully ... until now.  If you don't have time to look at this, I 
will try to understand it and fix it myself.  However, I really don't 
understand the code, and you certainly did when you wrote it ;-)


	  Thanks,
	  Spencer Graves


On 5/24/21 10:35 AM, Duncan Murdoch wrote:
> On 24/05/2021 10:17 a.m., Spencer Graves wrote:
>> Hi, Duncan:
>>
>>
>>       Thanks again.
>>
>>
>>       I found XQuartz-2.8.1.dmg and added the 5 lines you suggested
>> (modified to 2.8.1) to "R-CMD-check.yaml".  However, the "GitHub Action"
>> status symbols disappeared, so I deleted them.
>>
>>
>>       I copied the "CRAN" function from fda to sos and deleted the
>> reference to fda from sos.
>>
>>
>>       Now I'm getting, 'could not find function "CRAN"' in "processing
>> vignette 'sos.Rnw'", even though CRAN is in the NAMESPACE for sos.
>>
>>
>>       Suggestions?
> 
> This one is easy.  You use CRAN() before library(sos).
> 
>>
>>
>>       Thanks,
>>       Spencer
>>
>>
>> p.s.  To use Sys.getenv("NOT_CRAN"), I'd have to set the environment
>> variable "NOT_CRAN" on all the test platforms other than CRAN, and
>> that's a level of complexity I'd rather not get into.
> I wouldn't call it "complex", but it could be tedious.
> 
> Duncan Murdoch



More information about the R-package-devel mailing list