[BioC] 64-bit R - problem with affxparser

Kasper Daniel Hansen kasperdanielhansen at gmail.com
Tue Mar 23 16:32:32 CET 2010 

The authoritative place to look for information is the r-devel list,
particular the posts from R-core.  The most recent post about this
issue is the one Martin provided a link to, which contains the
following

"We do not anticipate any major changes to the build before 2.11.0 is
released (most likely in April), and it can now be regarded as at
'beta' status -- users with a 64-bit Windows OS and who do not need
Bioconductor packages are encouraged to try it out."

For specifics you should read the R-admin manual.  I am not sure if
the R-devel version of that manual is available anywhere on the net,
but if you have access to a R-devel build somewhere you will see it
has a section on building for 64 bit windows.  There is also a link to
a webpage with further comments.

But you will still need to install the latest version of the mingw64
toolchain.  I assume that eventually a snapshot will be included in
Rtools, but that does not seem to be the case yet.

Kasper

On Tue, Mar 23, 2010 at 3:57 AM, Michael Imbeault
<michael.imbeault at sympatico.ca> wrote:
> Hello,
>
> Sorry for the confusion about the two issues, I was under the impression
> that the build servers I linked to were the 64 bit version, my bad :)
>
> I only became aware of the affxparser on 64 bit problem yesterday, and
> stumbled upon the aroma.affymetrix group. Having not seen a post about it on
> the bioC list, I took the opportunity to post (and cc you). I'll try
> tomorrow to build 1.19.6 (I don't remember which build I tried to compile
> with mingw64).
>
> Regarding the instability of the keychain, do you happen to have a link
> where I could download the latest (semi) official version? There's a lot of
> uncertainty about this following various conversations on the web.
>
> Last, thanks for taking time to tackle this problem, and don't worry, I know
> that R 64 bit under Windows is kind of bleeding edge (and bioC running on it
> even more so) so I don't expect an overnight solution :)
>
> Cheers,
> Michael
>
> On 22/03/2010 11:20 PM, Kasper Daniel Hansen wrote:
>>
>> I was not aware of the discussion on the aroma.affymetrix group.  And
>> I was furthermore not aware of the fact that affxparser fails to build
>> at the moment (I thought we had automatic notification of this -
>> perhaps there is some issue with my spam filter?)  For another time,
>> it is usually productive to contact the maintainer :)
>>
>> As far as I can see the Bioconductor build the OP is referring to is
>> still using mingw32, which means that the build issue should be
>> completely unrelated to 64 bit windows (or the information at
>>
>> http://bioconductor.org/checkResults/2.6/bioc-LATEST/liverpool-NodeInfo.html
>> is outdated).
>>
>> I would expect affxparser to build with mingw64.  Or at least we
>> managed to build it (version 1.19.6 to be specific) under some version
>> of the toolchain (at that point the toolchain changed often I think)
>> with help from Prof. Ripley.  The toolchain has now stabilized and I
>> am in the process of getting a 64bit Windows 7 up and running so I can
>> make sure affxparser (and Rgraphviz - although initial results
>> indicate that Rgraphviz will be a pain) will work on this platform.
>>
>> That leaves the issue of why affxparser does not build on the
>> Bioconductor build servers.  I will look into this over the coming
>> days, but it will be much easier when I actually have a windows box to
>> test on.
>>
>> Finally, in general I suggest that people who want to use 64 bit
>> Windows arm themselves with some patience.  This is a new platform and
>> it is only very recently that the R windows team settled on which
>> compiler they (probably) will use and mingw64 is still somewhat
>> experimental (as I understand).  To top it off, not everyone has a 64
>> bit windows box lying around for testing, so it will still take a
>> while for things to be "normal".
>>
>> Kasper
>>
>> On Mon, Mar 22, 2010 at 7:45 PM, Martin Morgan<mtmorgan at fhcrc.org>  wrote:
>>
>>>
>>> Hi Michael --
>>>
>>> On 03/22/2010 03:59 PM, Michael Imbeault wrote:
>>>
>>>>
>>>> Hi,
>>>>
>>>> A toolchain to build bioC packages from source under R win 64 bit is
>>>> available at http://download.cos.name/incoming/Rtools64.zip  (see
>>>> https://stat.ethz.ch/pipermail/r-devel/2010-January/056280.html for the
>>>> original post by Gong Yu).
>>>>
>>>
>>> but this
>>>
>>> https://stat.ethz.ch/pipermail/r-devel/2010-March/056815.html
>>>
>>> suggests that the tool chain used at CRAN is different.
>>>
>>>
>>>>
>>>> However, some packages don't build/compile correctly, notably affxparser
>>>> (see the liverpool build / check for affxparser at
>>>> http://bioconductor.org/checkResults/2.6/bioc-LATEST/).
>>>>
>>>> This blocks the build of the aroma.affymetrix package under R win 64
>>>> (among others) as it depends on it - could we find the source of this
>>>> problem (I cc'd Dr. Hansen, the maintainer of the package)? This is a
>>>> problem for many users right now
>>>>
>>>> (http://groups.google.com/group/aroma-affymetrix/browse_thread/thread/1c01fdd3b243d3ec/5160fa633ea399a4?lnk=raot#5160fa633ea399a4).
>>>>
>>>
>>> I don't think Zaid or others on this thread communicated with the Bioc
>>> mailing list; maybe with Kasper directly? The hard-core Windows
>>> 64-bitters can check out affxparser 1.19.0 from svn (r42686; 1.19.0 did
>>> build successfully on Windows, and is the version that Windows
>>> biocLite() retrieves for win32). Which is not to say that Kasper won't
>>> address the Windows issues...
>>>
>>> Martin
>>>
>>>
>>>>
>>>> See
>>>>
>>>> http://bioconductor.org/checkResults/2.6/bioc-LATEST/affxparser/liverpool-checksrc.html
>>>> for a detailed view of the install.out log.
>>>>
>>>> Regards,
>>>> Michael
>>>>
>>>>
>>>> On 22/03/2010 5:11 PM, Martin Morgan wrote:
>>>>
>>>>>
>>>>> On 03/22/2010 12:28 PM, Jin, Jp wrote:
>>>>>
>>>>>
>>>>>>
>>>>>> Dear list,
>>>>>>
>>>>>> I would want to install R/bioconductor to my 64-bit machine with
>>>>>> Windows 7 installed. When are the 64-bit R build and bioconductor
>>>>>> available? will R 2.11 are Bio C 2.6  work with Windows 7?
>>>>>>
>>>>>>
>>>>>
>>>>> Hi --
>>>>>
>>>>> 32 bit R / Bioc installs on 64-bit windows 7.
>>>>>
>>>>> There is a 64-bit Windows R-devel; follow the 'windows64' link from a
>>>>> CRAN 'R binaries' page, e.g., at http://cran.fhcrc.org/bin/. There may
>>>>> be 64-bit Bioconductor packages in the next release but that is not
>>>>> certain. You will still be able to build your own 64-bit packages,
>>>>> although I am not exactly clear on details about the compiler required
>>>>> and availability of an appropriate Windows build tool chain.
>>>>>
>>>>> Martin
>>>>>
>>>>>
>>>>>
>>>>>>
>>>>>> Thanks,
>>>>>>
>>>>>> Jp Jin
>>>>>>
>>>>>> [[alternative HTML version deleted]]
>>>>>>
>>>>>> _______________________________________________ Bioconductor mailing
>>>>>> list Bioconductor at stat.math.ethz.ch
>>>>>> https://stat.ethz.ch/mailman/listinfo/bioconductor Search the
>>>>>> archives:
>>>>>> http://news.gmane.org/gmane.science.biology.informatics.conductor
>>>>>>
>>>>>>
>>>>>
>>>>>
>>>
>>> --
>>> Martin Morgan
>>> 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
>>>
>>> _______________________________________________
>>> Bioconductor mailing list
>>> Bioconductor at stat.math.ethz.ch
>>> https://stat.ethz.ch/mailman/listinfo/bioconductor
>>> Search the archives:
>>> http://news.gmane.org/gmane.science.biology.informatics.conductor
>>>
>>>
>>
>>
>
> _______________________________________________
> Bioconductor mailing list
> Bioconductor at stat.math.ethz.ch
> https://stat.ethz.ch/mailman/listinfo/bioconductor
> Search the archives:
> http://news.gmane.org/gmane.science.biology.informatics.conductor
>

More information about the Bioconductor mailing list