On Tue, 22 Jun 2010 04:30:12 +1000 (EST) Bruce Evans <b...@optusnet.com.au> wrote:
> On Mon, 21 Jun 2010, Alexander Kabaev wrote: > > > On Tue, 22 Jun 2010 03:22:40 +1000 (EST) > > Bruce Evans <b...@optusnet.com.au> wrote: > > <SKIP> > > > > DTrace _does_ try to unput EOF though and apparently gets away with > > it on Solaris, so while yor version is correct, it is also useless. > > Do you mean that it tries to unput EOF as an int (not obtained from a > char), and that that must fail and not unput ((unsigned char)EOF)? > Then the current version is still broken on platforms with chars > signed, since when it tries to unput a char with value EOF, that will > fail and not unput ((unsigned char)<char's value>). Correct, but chances of valid DTrace script having 0xff character in them is pretty much nonexistent, so I let that slide rather than pushing more changes into externally maintained code. To be absolutely correct, I think, we need to declare unput() as inline/static taking int argument and make sure that we invoke it as unput((unsigned char)*q) every time q is declared as char *. -- Alexander Kabaev
signature.asc
Description: PGP signature
