You could go down this rabbit hole if you like, but I suspect it has something 
to do with maintaining compatibility with multiple operating systems and I 
doubt R Core will explore the consequences of this change for you. Windows, for 
example, allows long file names but has some nasty corners that break if your  
total path+filename is 260+ characters and packages do get installed in 
directories fairly deep into the directory structure on a regular basis (e.g. 
packrat). Other tools that the R ecosystem depends on could also have internal 
constraints that come up when one tool builds on another and so on.

Have fun with that!

On April 18, 2019 11:02:15 AM PDT, Steven Scott <steve.the.bayes...@gmail.com> 
wrote:
>Thanks Dirk,
>Yes, I've done the same, and I agree the rules are the rules.  Rules
>should
>be updated when they're no longer helpful, or when their cost outweighs
>their benefit.  I'm curious whether that might be the case here.
>Cheers,
>Steve
>
>
>On Thu, Apr 18, 2019 at 10:54 AM Dirk Eddelbuettel <e...@debian.org>
>wrote:
>
>>
>> On 18 April 2019 at 10:37, Steven Scott wrote:
>> | My Boom package makes a C++ library available to package authors
>(mainly
>> | me).  The wrapped library is used outside of R and must comply with
>> | external style rules such as UseLongDescriptiveNames, and files
>must be
>> | named for the class they contain.  From time to time a
>> | LongDescriptiveFileName, when paired with its full directory path,
>> exceeds
>> | 100 characters.
>> |
>> | This creates a conflict with CRAN's rules about long file names,
>which
>> | stems from tar.  I'm wondering what this community thinks about
>asking
>> for
>> | that rule to be relaxed.  Both gnu tar and posix tar now allow
>unlimited
>> | length filenames, and the ustar format allows names up to 256
>characters.
>> |
>> | I'm interested in the opinions of people on this list about whether
>this
>> | rule has outlived its usefulness.  Thanks.
>>
>> There are no "opinions". There is CRAN Repo Policy.
>>
>> The BH package your Boom depends upon is actually named BH in part
>because
>> having a two-letter name shrunk the set of files violating this very
>> constraint.  Yet at every release I still get to renamed one file,
>and
>> update
>> one include statement.  All documented in the ChangeLog.
>>
>> So I would change the filenames, and move on.
>>
>> Dirk
>>
>> --
>> http://dirk.eddelbuettel.com | @eddelbuettel | e...@debian.org
>>
>
>       [[alternative HTML version deleted]]
>
>______________________________________________
>R-package-devel@r-project.org mailing list
>https://stat.ethz.ch/mailman/listinfo/r-package-devel

-- 
Sent from my phone. Please excuse my brevity.

______________________________________________
R-package-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-package-devel

Reply via email to