On Thu, 15 Mar 2007 19:29:41 +0100 Joerg.Schilling at fokus.fraunhofer.de (Joerg Schilling) wrote: > "William James" <williamjamesgnusolaris at gmail.com> wrote:
> > Why can't you just compile the ksh93 wordexp() for all libc versions? > > Or set an environment variable to tell applications to use the ksh93 > > code? > ksh93 uses sfio while libc uses stdio. > >From my understanding, libc would need to convert to libast in order to > be compatible with a wordexp() implementation that directly uses ksh93 code. the sh spec does not require that the shell use stdio is there any expectation that a call to wordexp() would/could/should modify the stdin/stdout/stderr buffer state of the caller at all? if a call to wordexp() does not touch the caller's stdio state why would it matter what io mechanism is used? the wordexp() implementation is free to use raw fd io, no? assuming proper cleanup on woreexp/wordfree return, how could the caller know that something other than stdio was used? -- Glenn Fowler -- AT&T Research, Florham Park NJ --
