On Dec 12 09:44, Ken Brown wrote: > On 12/12/2008 9:31 AM, Corinna Vinschen wrote: >> On Dec 11 13:14, Ken Brown wrote: >>> >>> I think you forgot to mention the change to st_nlink [1], which will >>> affect some applications. For example, the kpathsea search library for >>> texlive needs to be patched to accommodate the change. (I've sent the >>> patch to the tlbuild list in case anyone is interested.) This might also >>> apply to cygwin's tetex package, but I haven't tested it. >> Yes, that's right. I forgot about that. I didn't know that more than >> just findutils and coreutils are affected. In theory, every tool >> should be able to deal with st_nlink == 1. It's not the right value >> for an emtpy dir, which would be 2, so any tool should check what's >> actually going on in that dir. > > In the case of kpathsea, the program was deliberately relying on st_nlink > to speed up searching on systems that it viewed as "unix-like", including > cygwin. [It didn't check for subdirectories unless st_nlink > 2.] > Fortunately, this was documented, and it was easy to make a configuration > change to tell it to exclude cygwin for this purpose. I don't know how > likely it is that other applications will similarly cut corners.
I don't think it's the right way to special case this for Cygwin in kpathsea. This isn't exactly a Cygwin issue. The right thing to do would be to change the test in a generic fashion from if (st_nlink > 2) to if (st_nlink != 2) Corinna -- Corinna Vinschen Please, send mails regarding Cygwin to Cygwin Project Co-Leader cygwin AT cygwin DOT com Red Hat -- Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple Problem reports: http://cygwin.com/problems.html Documentation: http://cygwin.com/docs.html FAQ: http://cygwin.com/faq/
