[Rd] Wrong/incomplete documentation regarding the 100-bytes path length limit.

Tomas Kalibera tom@@@k@||ber@ @end|ng |rom gm@||@com
Tue Sep 17 14:45:01 CEST 2024


On 9/17/24 12:28, Iago Bonnici via R-devel wrote:
> Hello @r-devel,
>
>       From what I read in the following thread:
>
> https://stat.ethz.ch/pipermail/r-package-devel/2023q4/010197.html
>
>       The major reason to limit paths lengths in R packages is to help
> support MS Windows 260-chars paths lengths limit (pre-W10).
Not really, the limit on Windows is a reason why path length is a scarce 
resource in portable applications. And it exists also on Windows 10 and 
newer - it is a limitation now in Windows applications, inherited by a 
limit in early versions of Windows operating system. R itself doesn't 
have this limitation anymore.
>       However, the official documentation at this address:
>
> https://cran.r-project.org/doc/manuals/R-exts.html#Package-structure-1
>
>       States:
>
>   > /packages are normally distributed as tarballs, and these have a
> limit on path lengths: for maximal portability 100 bytes.
>
> /    This is the wrong explanation. Tar has been supporting 256-chars
> paths since POSIX.1-1988 and unlimited paths since POSIX.1-2001. The
> limitation should be deferred to MS Windows instead of tar.
>
>       Considering that the manuals are « /edited by the R Development
> Core Team /» <https://cran.r-project.org/manuals.html>, I suppose that
> this is the right list to point out this inconsistency? How can I help
> in fixing this? Should I open a bug at R Bugzilla?
>
>       Note: I am not discussing whether the limit itself should be
> reconsidered. That discussion seems yet covered by the aforementioned
> thread (but it's non-official), and by bug #18637
> <https://bugs.r-project.org/show_bug.cgi?id=18637> (but it's stale). In
> any case, I think there is no blocker to clarifying the situation in the
> documentation.

I have added a sentence to R-exts mentioning Windows is also a reason 
why it makes sense to keep paths short.

Best
Tomas

>
>       Best regards,
>
> --
> Iago-lito
> 	[[alternative HTML version deleted]]
>
> ______________________________________________
> R-devel using r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-devel



More information about the R-devel mailing list