[BioC] SVGAnnotation

[Uwe Ligges]

On 05.11.2010 20:49, Hervé Pagès wrote:
> On 11/05/2010 12:29 PM, Uwe Ligges wrote:
>>
>>
>> On 05.11.2010 20:25, Hervé Pagès wrote:
>>> Hi Uwe,
>>>
>>> On 11/05/2010 06:04 AM, Uwe Ligges wrote:
>>>> Hi,
>>>>
>>>> the binary of SVGAnnotation in
>>>> BioC/packages/2.7/extra/bin/windows/contrib/2.12/SVGAnnotation_0.6-0.zip
>>>>
>>>>
>>>> is declared to be built under R-2.10.x (rather than R-2.12.x)
>>>
>>> Thanks for telling us!
>>>
>>> We host a few Omegahat packages in our extra repo and we try to
>>> host the latest versions. This SVGAnnotation Windows binary is
>>> the latest version currently available at Omegahat.
>>>
>>> Unfortunately Omegahat binary packages are often lagging behind
>>> the source packages (when they are available at all). For example,
>>> for the current release, there is only 1 Windows binary at the
>>> moment (out of 84 source packages):
>>>
>>> http://www.omegahat.org/R/bin/windows/contrib/2.12/PACKAGES
>>>
>>> In the case of SVGAnnotation and XMLSchema (the Windows binary
>>> of XMLSchema that we host is also outdated), they don't contain
>>> native code so I should be able to produce updated binaries without
>>> too many problems.
>>
>> I see.
>>
>> A recent XMLSchema binary is on CRAN extras:
>> http://www.stats.ox.ac.uk/pub/RWin/bin/windows/contrib/2.12/
>>
>>
>>>>
>>>> which causes update.packages(checkBuilt=TRUE) to reinstall it again and
>>>> again under Windows.
>>>
>>> It sounds like update.packages(checkBuilt=TRUE) could do a better job.
>>> Generally speaking why would an "update tool" replace something by
>>> something that is identical?
>>
>>
>> Well, since update.packages assumes that packages in ./2.12/ have been
>> built with R-2.12.x which is not that unreasonable. ;-)
>
> Yes it's a fair assumption. Even though sometimes it's not the case.
> Beside us dropping old binaries in our ./2.12/, the CRAN repository
> currently has a symlink from 2.13 to 2.12 so right now packages in
> ./2.13/ have not been built with R-2.13.x
> I wonder what update.packages(checkBuilt=TRUE) would do for R-devel
> users on Windows (I didn't try)...


Good, point. It does reinstall the packages again and again. But things 
are again diferent for R-devel: It may be the case that a package built 
under a certain svn revision does not work with the next svn revision of 
R, hence I think it is not the best way but OK to keep R-devel separate.

Uwe







> H.
>
>
>> If it is expected to do a better job, we'd need to add the build stamp
>> into the PACKAGES database.
>>
>> Best wishes,
>> Uwe
>>
>>
>>
>>> Cheers,
>>> H.
>>>
>>>>
>>>> Best wishes,
>>>> Uwe
>>>>
>>>> _______________________________________________
>>>> 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
>>>
>>>
>
>