This is a narrow view of the world. As has been mentioned here by Tomas, the issue at this point is that a very widely-used operating system does not allow the absolute path to be longer than 254 characters unless users make possibly-breaking changes to their OS configuration. If a user is currently working in a directory with a path that is 150 characters long, and you try to un-tar a package there, it will work. If they are in a directory path that is 170 characters long, then the un-tar will fail. They then have to do some research and discover that they have to share the total absolute path length with the package creators, and they become accustomed to working in directories with path length no longer than 154 characters.
Now you want to propose giving the entire path length budget plus 2 to the package developers. Not gonna happen. And Dirk wants 50 more characters. Then the user working at 150 characters tries to un-tar a package with paths 150 characters long and 150 user plus 150 package gets you to 300 characters and what used to work for them stops working suddenly. I think a package developer can figure out how to shorten their internal paths easier than thousands of users can restructure their historically useful directory structures or break their existing software. I vote no to both of these proposals. Eventually, Microsoft is going to virtualize running old Windows APIs, and once people migrate away from the old API then this stupid problem will go away. But as it is we are stuck. On December 13, 2023 7:27:42 AM PST, "McGrath, Justin M" <jmcgr...@illinois.edu> wrote: >I'm not even asking for that. I'm asking for the decades old ustar >comparability, which is 255 characters, at most. In practice it is slightly >less than that. ustar is older than R itself. There is no one using R who >doesn't have ustar compatible tar. > >________________________________________ >From: Dirk Eddelbuettel <e...@debian.org> >Sent: Wednesday, December 13, 2023 8:59 AM >To: Tomas Kalibera >Cc: McGrath, Justin M; Ben Bolker; Martin Maechler; >r-package-devel@r-project.org >Subject: Re: [R-pkg-devel] Wrong mailing list: Could the 100 byte path length >limit be lifted? > > >On 13 December 2023 at 15:32, Tomas Kalibera wrote: >| Please don't forget about what has been correctly mentioned on this >| thread already: there is essentially a 260 character limit on Windows >| (see >| >https://urldefense.com/v3/__https://blog.r-project.org/2023/03/07/path-length-limit-on-windows/index.html__;!!DZ3fjg!-YZg5PVpulgNXCVfVklP442UG_0ofB8omMLlMq5dNadF9RP_6uofnrT6IbZpG1fpvjmAtLyAm1Y9rbc$ >| for more). Even if the relative path length limit for a CRAN package was >| no longer regarded important for tar compatibility, it would still make >| sense for compatibility with Windows. It may still be a good service to >| your users if you keep renaming the files to fit into that limit. > >So can lift the limit from 100 char to 260 char ? > >Dirk > >-- >dirk.eddelbuettel.com | @eddelbuettel | e...@debian.org > >______________________________________________ >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
