From: | "Devon H. O'Dell " <dodell@xxxxxxxxxxxxxxx> |
Date: | Sat, 2 Apr 2005 16:59:06 +0200 |
Mail-followup-to: | kernel@crater.dragonflybsd.org |
On Sat, Apr 02, 2005 at 04:47:28PM +0200, Joerg Sonnenberger wrote: > On Sat, Apr 02, 2005 at 04:05:50PM +0200, Devon H. O'Dell wrote: > > Hey, > > > > I dunno if there are other files that are hard linked together > > and how these are all testing what name they were called by, but > > Liam pointed out the case of rm(1) to me today and asked why it > > was going about that like it was. > > At least gzip and compress come to my mind directly. I'll see what I can do about these two as well. > > I also don't get it. > > > > To me, it seems that something like: > > > > if (strncmp("unlink", basename(*argv)) == 0) > > > > would make more sense than how it's doing it now (with a > > temporary pointer to const char and strrchr). I know that > > basename(3) isn't POSIX -- is this an issue? > > Nope, the problem is that basename(3) has a different semantic. > It does strip trailing slashes, which should not happen here. > The easiest and fastest way is calling getprogname(), crt1.c already > did the necessary parsing (with a poor man's version of strrchr BTW). > > Joerg Why should that `not happen here'? Neither unlink nor rm should be called as unlink////. Perhaps I'm misunderstanding. I'll take a look at getprogname. --Devon
Attachment:
pgp00001.pgp
Description: PGP signature