[Bioc-devel] os x binary

[Kasper Daniel Hansen]

Dan

On Snow Leopard I installed the CRAN binary of R.  I made sure it
overwrote whatever I had compiled from source and I uninstalled all
packages from my self-compiled installation, leaving me with a
standard R-2.14 installation.

If I then installed affxparser using type = "mac.binary.leopard" and
tried readCelHeader on the "v5array1.CEL" R crashed with
  "abort trap"

If I subsequently installed affxparser using type = "source" it worked.

I then did a svn checkout of the 2.9 release branch of affxparser
giving me 1.26.2.  I produced a binary tarball by
  R CMD build affxparser
  R CMD INSTALL --binary affxparser_1.26.2.tar.gz

I will send both in a separate email.  I tested the binary by
installing into the CRAN binary and confirmed that it worked by
running the test above.

My C compiler details are (affxparser is mostly C++, no fortran)

sleipner-2:~/Work/packages/tmp/> R CMD config CC
gcc-4.2 -arch x86_64 -std=gnu99

sleipner-2:~/Work/packages/tmp/> `R CMD config CC` --version
i686-apple-darwin10-gcc-4.2.1 (GCC) 4.2.1 (Apple Inc. build 5666) (dot 3)
Copyright (C) 2007 Free Software Foundation, Inc.
This is free software; see the source for copying conditions.  There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

sleipner-2:~/Work/packages/tmp/> R CMD config CXX
g++-4.2 -arch x86_64

sleipner-2:~/Work/packages/tmp/> `R CMD config CXX` --version
i686-apple-darwin10-g++-4.2.1 (GCC) 4.2.1 (Apple Inc. build 5666) (dot 3)
Copyright (C) 2007 Free Software Foundation, Inc.
This is free software; see the source for copying conditions.  There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

So there seems to be something wrong with the Bioconductor build.

Kasper


On Thu, Nov 17, 2011 at 7:04 PM, Dan Tenenbaum <dtenenba at fhcrc.org> wrote:
> On Thu, Nov 17, 2011 at 3:53 PM, Kasper Daniel Hansen
> <kasperdanielhansen at gmail.com> wrote:
>> On Thu, Nov 17, 2011 at 5:27 PM, Dan Tenenbaum <dtenenba at fhcrc.org> wrote:
>>> On Thu, Nov 17, 2011 at 2:18 PM, Kasper Daniel Hansen
>>> <kasperdanielhansen at gmail.com> wrote:
>>>> On Thu, Nov 17, 2011 at 5:02 PM, Dan Tenenbaum <dtenenba at fhcrc.org> wrote:
>>>>> Hi Kasper,
>>>>>
>>>>> On Thu, Nov 17, 2011 at 1:47 PM, Kasper Daniel Hansen
>>>>> <kasperdanielhansen at gmail.com> wrote:
>>>>>> Hi Dan
>>>>>>
>>>>>> I am on snow leopard.  We have not made any changes to the C code in
>>>>>> affxparser in over a year I think, and it works on a wide variety of
>>>>>> systems.  Everyone who has reported this problem seems to be able to
>>>>>> fix it by re-installing the CRAN binary and rebuilding affxparser from
>>>>>> source.  I am not sure that this includes anyone on Lion.
>>>>>>
>>>>>> Have you manually rebuild affxparser?  We have not bumped the version
>>>>>> number since we have made no changes to the package, so according to
>>>>>> my understanding of the build system it will not get rebuild
>>>>>> automatically.
>>>>>
>>>>> I bumped the version number yesterday to 1.26.2, and that is what was
>>>>> built by our build system and is now available via biocLite().
>>>>>
>>>>> If I build affxparser from source on my Lion machine, readCdfHeader() works.
>>>>
>>>> So the binary version works on Leopard but not on Lion (if I
>>>> understand you correctly).  Building from source works on Lion.
>>>> Doesn't this tell us that there is something wrong with the binaries?
>>>
>>> Yes, your understanding is correct.
>>>
>>> It does seem to tell us that there is a problem with the binaries, at
>>> least the one produced on Leopard (which is what our build system
>>> uses).
>>>
>>> Earlier in this thread, you said installing the latest R 2.14.0 binary
>>> from CRAN would solve this problem. It appears that it does not solve
>>> it, at least for some machines (Lion may or may not be the deciding
>>> factor).
>>
>> Well, actually my statement is not based on my own testing.  But this
>> is what I have told people who had problems and it seems that made the
>> problem go away.  I am building R from source myself, so I am not
>> using the binary.
>
> Maybe we should clarify which R binary we are talking about? On our
> build system for release we use R 2.14.0 installed from this package:
>
> http://cran.r-project.org/bin/macosx/R-2.14.0.pkg
>
> "curl -I" reports that this file has not been changed since October 31.
>
> So I'm not clear on what Simon actually rebuilt. I looked on R-sig-mac
> and the only reference he makes to rebuilding (that I could find) is
> something to do with R.app, and whatever binary he rebuilt is probably
> on his att.com site, because it's not on CRAN.
>
>
>
>>
>> Perhaps Oliver can say exactly what he did to get his problem solved?
>>
>
> That would be great. He said he built from source, but it would be
> great to see sessionInfo(), etc.
>
>>> If you want to send me (off-list) the .tgz file produced by
>>> R CMD INSTALL --build affxparser
>>> on your Snow Leopard system, I'll test if it works on my Lion system.
>>
>> On its way.
>>
>
> Thanks
> Dan
>
>>>
>>> Dan
>>>
>>>
>>>
>>>>
>>>> Kasper
>>>>
>>>>>
>>>>> Dan
>>>>>
>>>>>
>>>>>>
>>>>>> Kasper
>>>>>>
>>>>>> On Thu, Nov 17, 2011 at 4:01 PM, Dan Tenenbaum <dtenenba at fhcrc.org> wrote:
>>>>>>> Hi Kasper,
>>>>>>>
>>>>>>> On Mon, Nov 14, 2011 at 1:24 PM, Dan Tenenbaum <dtenenba at fhcrc.org> wrote:
>>>>>>>> On Mon, Nov 14, 2011 at 12:57 PM, Kasper Daniel Hansen
>>>>>>>> <kasperdanielhansen at gmail.com> wrote:
>>>>>>>>> Some problems with the CRAN binary for OS X has been reported on
>>>>>>>>> r-sig-mac.  Oliver (cc'ed) reported a problem with affxparser
>>>>>>>>> segfaulting that we solved by
>>>>>>>>> (1) installing an updated R binary
>>>>>>>>> (2) recompiling affxparser from source
>>>>>>>>>
>>>>>>>>> It was not enough to do (1) and the reinstall using biocLite and type
>>>>>>>>> = "mac.binary".
>>>>>>>>>
>>>>>>>>> The building team might already be aware of this, but in case they are
>>>>>>>>> not I would suggest re-installing the R binary and rebuilding
>>>>>>>>> packages.
>>>>>>>>>
>>>>>>>>
>>>>>>>> Thanks, I will update the R on pitt (Mac machine used for release
>>>>>>>> builds) as soon as it is finished with the builds it's running now.
>>>>>>>>
>>>>>>>
>>>>>>> The new affxparser produced by the latest R binary is now available,
>>>>>>> and on my machine (running Lion) it still produces the same segfault:
>>>>>>>
>>>>>>> R --vanilla -q -e 'library(affxparser);sessionInfo(); h <-
>>>>>>> readCdfHeader("Mapping250K_Nsp.cdf")'
>>>>>>>> library(affxparser);sessionInfo(); h <- readCdfHeader("Mapping250K_Nsp.cdf")
>>>>>>> R version 2.14.0 (2011-10-31)
>>>>>>> Platform: x86_64-apple-darwin9.8.0/x86_64 (64-bit)
>>>>>>>
>>>>>>> locale:
>>>>>>> [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
>>>>>>>
>>>>>>> attached base packages:
>>>>>>> [1] stats     graphics  grDevices utils     datasets  methods   base
>>>>>>>
>>>>>>> other attached packages:
>>>>>>> [1] affxparser_1.26.2
>>>>>>> Abort trap: 6
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> On a different machine (running Leopard) it works:
>>>>>>>
>>>>>>> compbio$ R --vanilla -q -e 'library(affxparser);sessionInfo(>
>>>>>>>> library(affxparser);sessionInfo(); h <- readCdfHeader("Mapping250K_Nsp.cdf")
>>>>>>> R version 2.14.0 (2011-10-31)
>>>>>>> Platform: i386-apple-darwin9.8.0/i386 (32-bit)
>>>>>>>
>>>>>>> locale:
>>>>>>> [1] C
>>>>>>>
>>>>>>> attached base packages:
>>>>>>> [1] stats     graphics  grDevices utils     datasets  methods   base
>>>>>>>
>>>>>>> other attached packages:
>>>>>>> [1] affxparser_1.26.2
>>>>>>>>
>>>>>>>>
>>>>>>>
>>>>>>> So I think there may still be a problem.
>>>>>>> Dan
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>> Dan
>>>>>>>>
>>>>>>>>
>>>>>>>>> Kasper
>>>>>>>>>
>>>>>>>>> _______________________________________________
>>>>>>>>> Bioc-devel at r-project.org mailing list
>>>>>>>>> https://stat.ethz.ch/mailman/listinfo/bioc-devel
>>>>>>>>>
>>>>>>>>
>>>>>>>
>>>>>>
>>>>>
>>>>
>>>
>>
>