Stefano,
I was looking at some posts with regards to Tiger and R. Could you
tell me what is going on with the 64 bit version? Would this be
compatible with the Bioconductor Binaries?
Much thanks for you and your groups efforts, R is a remarkable
package.
David Seo
64-bits aside, I'd appreciate an update on what the community
consensus is regarding Tiger and BioC. Now that I've managed to build
R for Tiger, I'm trying to install various BioC packages and running
into trouble. Two initial problems:
1. gcrma can't seem to find the R framework (granted, this is
installed in a non-standard location, but other packages seem to be
able to do this fine)
2. SparseM doesn't seem to like the gfortran compiler.
3. affyPLM has a log2 function which conflicts with the defintion in
math.h. In particular, the LESN.c version is declared static and the
math.h one isn't. GCC 4 chokes on this.
I'm sure there's more, but this are the initial show-stoppers for me.
Thanks,
Cyrus
On May 4, 2005, at 10:56 AM, David Seo wrote:
> Stefano,
>
> I was looking at some posts with regards to Tiger and R. Could you
> tell me what is going on with the 64 bit version? Would this be
> compatible with the Bioconductor Binaries?
>
> Much thanks for you and your groups efforts, R is a remarkable
> package.
>
> David Seo
>
> _______________________________________________
> Bioconductor mailing list
> Bioconductor@stat.math.ethz.ch
> https://stat.ethz.ch/mailman/listinfo/bioconductor
>
Hi Robert,
Sorry for the lack of specifics. More details follow:
On May 6, 2005, at 3:06 PM, Robert Gentleman wrote:
> Please always provide information on the version of R, and the
> package versions - it is important for us in trying to help.
All of this pertains to r-devel (2.2.0 prerelease) from today's SVN
sources. Perhaps I'm too far out on the bleeding edge to get things
working properly with BioC, but I'd like to help get the latest R-
devel release and develOK versions of BioC packages working properly
on Tiger.
>> 1. gcrma can't seem to find the R framework (granted, this is
>> installed in a non-standard location, but other packages seem to
>> be able to do this fine)
> Can you be a bit more precise? What is the error message exactly?
I don't have the precise error message, but the problem was that I
had installed the R framework in a non-standard location and the -
framework R option in LIBR wasn't sufficient to find the framework.
I've since reported this to R-SIG-MAC as this is a general R problem,
not a bioconductor problem. The solution is to include a -F<path to="" the="" directory="" where="" the="" framework="" is="" installed=""> option in the LIBR
line in Makeconf. I've done this by hand and now I can build things
like gcrma. Some autoconf magic that is a bit over my head is
presumably required to do this properly.
>> 2. SparseM doesn't seem to like the gfortran compiler.
>
> That is really an issue for the maintainer of SparseM. We don't
> have the right to just start changing their code on them. Have you
> reported this, with complete output, to them? In my experience they
> are very receptive and quick to fix problems.
Thanks. It wasn't obvious to me that SparseM wasn't part of
Bioconductor (I'm using (trying to use it) it because gostats (I
think) wants it). I've also backed down to the f77 compiler which
avoids the problem for the moment.
>> 3. affyPLM has a log2 function which conflicts with the defintion
>> in math.h. In particular, the LESN.c version is declared static
>> and the math.h one isn't. GCC 4 chokes on this.
> This one should be fixable in the short term. We will see what
> we can do.
Great. I'm not sure if the solution is to call the systems log2 or to
rename the LESN.c log2 function, but it looks like this is a name
collision in a language that doesn't have good namespace support.
Thanks for addressing this.
> Again, thanks for the report - we will see what we can do
Thank you for the prompt response and for pointing me in the proper
directions!
Sincerely,
Cyrus
It would be nice to work with R-2.1.x for the moment.
BTW, we are committed to deliver a tiger build of R and I'm going to
setup a machine for daily building of packages of both CRAN and BioC.
There are still minor issues to fix (have a look at last threads on
R-sig-Mac).
stefano
On 07/mag/05, at 00:18, Cyrus Harmon wrote:
>
> Hi Robert,
>
> Sorry for the lack of specifics. More details follow:
>
> On May 6, 2005, at 3:06 PM, Robert Gentleman wrote:
>
>> Please always provide information on the version of R, and the
>> package versions - it is important for us in trying to help.
>
> All of this pertains to r-devel (2.2.0 prerelease) from today's SVN
> sources. Perhaps I'm too far out on the bleeding edge to get things
> working properly with BioC, but I'd like to help get the latest
> R-devel release and develOK versions of BioC packages working
properly
> on Tiger.
>
>>> 1. gcrma can't seem to find the R framework (granted, this is
>>> installed in a non-standard location, but other packages seem to
be
>>> able to do this fine)
>> Can you be a bit more precise? What is the error message exactly?
>
> I don't have the precise error message, but the problem was that I
had
> installed the R framework in a non-standard location and the
> -framework R option in LIBR wasn't sufficient to find the framework.
> I've since reported this to R-SIG-MAC as this is a general R
problem,
> not a bioconductor problem. The solution is to include a -F the directory where the framework is installed> option in the LIBR
> line in Makeconf. I've done this by hand and now I can build things
> like gcrma. Some autoconf magic that is a bit over my head is
> presumably required to do this properly.
>
>
>>> 2. SparseM doesn't seem to like the gfortran compiler.
>>
>> That is really an issue for the maintainer of SparseM. We don't
>> have the right to just start changing their code on them. Have you
>> reported this, with complete output, to them? In my experience they
>> are very receptive and quick to fix problems.
>
> Thanks. It wasn't obvious to me that SparseM wasn't part of
> Bioconductor (I'm using (trying to use it) it because gostats (I
> think) wants it). I've also backed down to the f77 compiler which
> avoids the problem for the moment.
>
>
>>> 3. affyPLM has a log2 function which conflicts with the defintion
in
>>> math.h. In particular, the LESN.c version is declared static and
the
>>> math.h one isn't. GCC 4 chokes on this.
>> This one should be fixable in the short term. We will see what
we
>> can do.
>
> Great. I'm not sure if the solution is to call the systems log2 or
to
> rename the LESN.c log2 function, but it looks like this is a name
> collision in a language that doesn't have good namespace support.
> Thanks for addressing this.
>
>
>> Again, thanks for the report - we will see what we can do
>
> Thank you for the prompt response and for pointing me in the proper
> directions!
>
> Sincerely,
>
> Cyrus
>
> _______________________________________________
> Bioconductor mailing list
> Bioconductor@stat.math.ethz.ch
> https://stat.ethz.ch/mailman/listinfo/bioconductor
>
On 04/mag/05, at 19:56, David Seo wrote:
> Stefano,
>
> I was looking at some posts with regards to Tiger and R. Could you
> tell me what is going on with the 64 bit version? Would this be
> compatible with the Bioconductor Binaries?
a tiger + 64bit version will require a complete rebuild of all the
packages (at least the ones that contain C or Fortran code).
The 64bit is a general issue in R (see Brian Ripley's note on
developer.r-project.org).
stefano
>
> Much thanks for you and your groups efforts, R is a remarkable
package.
>
> David Seo
>
> _______________________________________________
> Bioconductor mailing list
> Bioconductor@stat.math.ethz.ch
> https://stat.ethz.ch/mailman/listinfo/bioconductor
>
