Re: [gentoo-user] Doubt about R support in gentoo

[Zhu Sha Zang]

Yeah, thank you. ICU is a great pain in the ring!


Regards


On 11/19/2016 12:26 PM, c0a800ff@×××××.com wrote:
> On 11/19/16 12:50, Alec Ten Harmsel wrote:
>>
>>
>> El 19/11/2016 a las 07:47 a. m., Zhu Sha Zang escribió:
>>>
>>> Hello everybody...
>>>
>>>
>>> A single doubt here:
>>>
>>>
>>> After an upgrade in package dev-libs/icu (from 57.1 to 58.1-r1) i
>>> tried a package upgrade inside R and got that:
>>>
>>>
>>> /> update.packages(checkBuilt=TRUE, ask=FALSE,
>>> repos='http://cran.us.r-project.org')
>>> trying URL 'http://cran.us.r-project.org/src/contrib/knitr_1.15.tar.gz'
>>> Content type 'application/x-gzip' length 1027358 bytes (1003 KB)
>>> ==================================================
>>> downloaded 1003 KB
>>>
>>> * installing *source* package ‘knitr’ ...
>>> ** package ‘knitr’ successfully unpacked and MD5 sums checked
>>> ** R
>>> ** demo
>>> ** inst
>>> ** preparing package for lazy loading
>>> Error in dyn.load(file, DLLpath = DLLpath, ...) :
>>>   *unable to load shared object
>>> '/usr/lib64/R/library/stringi/libs/stringi.so':**
>>> **  libicui18n.so.57: cannot open shared object file: No such file or
>>> directory*
>>> /
>>>
>>
>> I am not super familiar with R - can you try rebuilding R itself before
>> doing the CRAN package installs/upgrades? It looks like the path to ICU
>> is hard-coded somehow.
>>
>>>
>>> This kind of "bug" can I put inside bugs.gentoo.org or its a cran
>>> package bug?
>>>
>>> It is not the first time and i'm in dout about how to proceed.
>>>
>>>
>>
>> If rebuildling R fixes this problem, you can file a bug so that
>> developers somehow trigger an R rebuild whenever ICU is updated.
>>
>> Alec
>>
>
> Some R package update errors may be solved by creating a temporary
> folder, for example ~/R/tmp (placing it under ~/R has the advantage of
> sharing the same parent directory as the default per-user package
> library), and then passing the following environment variable to R:
> TMPDIR=$HOME/R/tmp. This allows certain configure scripts to be
> executed if your /tmp is mounted noexec.
>
> As for the stringi error, my system doesn't have it in the system-wide
> library, either:
>
> $ ls /usr/lib64/R/library/
> KernSmooth      compiler        mgcv            stats4
> MASS            datasets        nlme            survival
> Matrix          foreign         nnet            tcltk
> base            grDevices       parallel        tools
> boot            graphics        rpart           translations
> class           grid            spatial         utils
> cluster         lattice         splines
> codetools       methods         stats
>
> But:
> $ ls ~/R/x86_64-pc-linux-gnu-library/3.2/stringi
> AUTHORS         LICENSE         R               libs
> CITATION        Meta            help
> DESCRIPTION     NAMESPACE       html
> INDEX           NEWS            include
>
> Another possibility may be that your R has a useflag for icu (mine
> doesn't):
>
> $ equery uses R
> [ Legend : U - final flag setting for installation]
> [        : I - package is installed with flag     ]
> [ Colors : set, unset                             ]
>  * Found these USE flags for dev-lang/R-3.2.2:
>  U I
>      ...
>  + + doc         : Add extra documentation (API, Javadoc, etc). It is
>                    recommended to enable per package instead of globally
>  - - icu         : Enable ICU (Internationalization Components for
>                    Unicode) support, using dev-libs/icu
>  - - java        : Add support for Java
>      ...
>
>
> Vitor
>