[Bioc-devel] BioC 3.0 CHECK ERROR

[Martin Morgan]

On 11/26/2014 07:36 PM, Dan Tenenbaum wrote:
>
>
> ----- Original Message -----
>> From: "Martin Morgan" <mtmorgan at fredhutch.org>
>> To: "Dan Tenenbaum" <dtenenba at fredhutch.org>
>> Cc: "Joachim Schumann" <joachim.schumann at ufz.de>, bioc-devel at r-project.org
>> Sent: Wednesday, November 26, 2014 5:33:22 PM
>> Subject: Re: [Bioc-devel] BioC 3.0 CHECK ERROR
>>
>> On 11/26/2014 10:03 AM, Dan Tenenbaum wrote:
>>>
>>>
>>> ----- Original Message -----
>>>> From: "Martin Morgan" <mtmorgan at fredhutch.org>
>>>> To: "Joachim Schumann" <joachim.schumann at ufz.de>,
>>>> bioc-devel at r-project.org
>>>> Sent: Wednesday, November 26, 2014 9:00:07 AM
>>>> Subject: Re: [Bioc-devel] BioC 3.0 CHECK ERROR
>>>>
>>>> On 11/26/2014 01:39 AM, Joachim Schumann wrote:
>>>>> Hi everyone,
>>>>> this morning I got an email from bioconductor.
>>>>> The message:
>>>>>
>>>>> According to the "Build/check report" for BioC 3.0,
>>>>> the flowCHIC package has the following problem(s):
>>>>> ERROR for 'R CMD check' on zin1.
>>>>>
>>>>> The error I get is the following:
>>>>>
>>>>> Error in callGeneric() :
>>>>>      'callGeneric' with a primitive needs explicit arguments (no
>>>>>      formal args defined)
>>>>> Calls: calculate_overlaps_xor ... calculate_overlaps_xor ->
>>>>> .local
>>>>> -> bXOR ->
>>>>> round -> round -> callGeneric
>>>>>
>>>>> Does anyone know what the problem is? It seems that the round
>>>>> function causes
>>>>> the error.
>>>>
>>>> This is some kind of interaction with Matrix
>>>>
>>>>    > library(EBImage)
>>>>    > img = readImage(system.file(package="EBImage", "images",
>>>>    > "nuclei.tif"))
>>>>    > xx = round(img)
>>>>    > library(Matrix)
>>>>    > yy = round(img)
>>>> Error in callGeneric() :
>>>>      'callGeneric' with a primitive needs explicit arguments (no
>>>>      formal
>>>>      args defined)
>>>>
>>>> which defines a Math2 group generic. How long has this been
>>>> appearing? It
>>>> doesn't seem like the build system R, EBImage or Matrix has
>>>> changed
>>>> since
>>>> release, and the package is available via biocLite implying that
>>>> it
>>>> once built...
>>>
>>> I think the build system R has changed since release, since it's
>>> running 3.1.2 which was released on 10/31/2014, but release was
>>> 10/14/2014.
>>>
>>> Indeed, if I look at the flowCHIC_1.0.0.tgz DESCRIPTION file, I see
>>> in the "Built" line that it was built with R 3.1.1.
>>>
>>> So this suggests a problem with R-3.1.2? Is this a bug we should
>>> report?
>>
>> The path to Matrix is
>>
>> flowCHIC
>>     Imports: vegan
>>       Imports: mgcv
>>         Imports: Matrix
>>
>> vegan was updated on
>>
>>   > packageDescription("vegan")[c('Version', 'Date/Publication')]
>> $Version
>> [1] "2.2-0"
>>
>> $`Date/Publication`
>> [1] "2014-11-17 11:35:34"
>>
>> and changed (based on comparison with the last archived version,
>> 2.0-10 from
>> http://cran.fhcrc.org/src/contrib/Archive/vegan) to move Matrix from
>> Suggests:
>> to Imports:.
>>
>> I think it's a long-standing issue in R; I can reproduce it using the
>> EBImage /
>> Matrix example in R 2.15, for instance (although oddly I had to go
>> through some
>> mild histrionics to get EBImage to compile). Specifically, it seems
>> like a patch
>> that Michael just applied in R-devel (!), plus
>>
>> Index: MethodsListClass.R
>> ===================================================================
>> --- MethodsListClass.R	(revision 67060)
>> +++ MethodsListClass.R	(working copy)
>> @@ -353,7 +353,7 @@
>>                  function(x, digits) {
>>                      value <- x
>>                      x <- x at .Data
>> -                  value at Data  <- callGeneric()
>> +                  value at .Data  <- callGeneric()
>>                      value
>>                  })
>>        ## some methods for nonStructure, ensuring that the class and
>>        slots
>>
>>
>
> Are you going to commit this? If this (and Michael's change) fixes the problem, then I will update the build machines. BTW flowCHIC has the same issue in release, so I guess some backporting is in order.
>

Both R-devel and R-3-1-branch will have a fix.

For flowCHIC, I'd leave devel as-is, it will eventually heal itself.

For release, I think Joachim should explore an alternative so that users of the 
current R release can use flowCHIC without waiting for the next release in the 
R-3-1 series. I'm not really familiar with the EBImage package, but I think what 
you want to do is the equivalent of

   x = initialize(x, round(imageData(x) * (2^8 - 1), digits = 3))

i.e., calculating round() on the raw array, then updating the Image object 'x'.

Martin

> Dan
>
>
>> Martin
>>
>>>
>>> Dan
>>>
>>>
>>>>
>>>> Martin
>>>>
>>>>>
>>>>> Best,
>>>>> Joachim
>>>>>
>>>>
>>>>
>>>> --
>>>> Computational Biology / Fred Hutchinson Cancer Research Center
>>>> 1100 Fairview Ave. N.
>>>> PO Box 19024 Seattle, WA 98109
>>>>
>>>> Location: Arnold Building M1 B861
>>>> Phone: (206) 667-2793
>>>>
>>>> _______________________________________________
>>>> Bioc-devel at r-project.org mailing list
>>>> https://stat.ethz.ch/mailman/listinfo/bioc-devel
>>>>
>>
>>
>> --
>> Computational Biology / Fred Hutchinson Cancer Research Center
>> 1100 Fairview Ave. N.
>> PO Box 19024 Seattle, WA 98109
>>
>> Location: Arnold Building M1 B861
>> Phone: (206) 667-2793
>>


-- 
Computational Biology / Fred Hutchinson Cancer Research Center
1100 Fairview Ave. N.
PO Box 19024 Seattle, WA 98109

Location: Arnold Building M1 B861
Phone: (206) 667-2793