On Feb 23 10:25, Richard Elberger wrote: > I think I did more than I needed to. I extracted the inst package as > well. > > 64-bit. yeah I did the stupid thing of putting in x86 and ash’s > inability to run (for rebase) let me know that I had screwed something > up. > > I will follow your instruction below. My fault for not backing up > files beforehand - lesson learned. > > You may hear from me again soon - I have some weird problem with perl > -w (other tests failing in core). I need to use the ‘filetest > “strict”’ pragma in order to get it to work since that changes > behavior to include ACL evaluation. Basically it’s not treating > someone in the Administrators group as a ‘rootish’ user. i.e. if a > file is 0444 and user A is in Administrators group, -w test fails. > But under ‘filetest “strict”’ it passes. The queer thing here is that > in perl, it seems I can open the file for writing even if -w fails. > Anyways, it’s still not clear enough in my head to figure out if it’s > cygwin or perl that needs to account for it.
This is a known perl problem, afaik, and not reported for the first time. It seems the -r/-w/-x tests check the file permissions returned by stat(2), but they should test access based on the return value of access(2) or faccessat(2) calls instead. Corinna -- Corinna Vinschen Please, send mails regarding Cygwin to Cygwin Maintainer cygwin AT cygwin DOT com Red Hat
pgpbuXWnvb38_.pgp
Description: PGP signature
