PATCH: 3.1.5-pws-12: getcwd

PATCH: 3.1.5-pws-12: getcwd

[Peter Stephenson]

Another configure cock-up I made.  For the last couple of weeks everybody's
been using the native getcwd() on every architecture.  The fact nobody's
complained may be significant.

--- configure.in.cwd	Mon Mar 15 09:57:52 1999
+++ configure.in	Mon Mar 15 12:11:29 1999
@@ -847,8 +847,8 @@
 AC_CACHE_CHECK(whether we should use the native getcwd,
 zsh_cv_use_getcwd,
 [case "${host_cpu}-${host_vendor}-${host_os}" in
-    *QNX*) zsh_cv_use_getcwd=no ;;
-    *) zsh_cv_use_getcwd=yes ;;
+    *QNX*) zsh_cv_use_getcwd=yes ;;
+    *) zsh_cv_use_getcwd=no ;;
  esac])
 if test $zsh_cv_use_getcwd = yes; then
   AC_DEFINE(USE_GETCWD)

-- 
Peter Stephenson <pws@ibmth.df.unipi.it>       Tel: +39 050 844536
WWW:  http://www.ifh.de/~pws/
Dipartimento di Fisica, Via Buonarroti 2, 56127 Pisa, Italy

RE: PATCH: 3.1.5-pws-12: getcwd

[Andrej Borsenkow]

>
> Another configure cock-up I made.  For the last couple of weeks
> everybody's
> been using the native getcwd() on every architecture.  The fact nobody's
> complained may be significant.
>

What is wrong with native getcwd? (Apart, that it may use native malloc() to
allocate buffer)? If there are any known problems, is it possible to write a
configure check for them?

/andrej

Re: PATCH: 3.1.5-pws-12: getcwd

[Peter Stephenson]

"Andrej Borsenkow" wrote:
> What is wrong with native getcwd? (Apart, that it may use native malloc() to
> allocate buffer)? If there are any known problems, is it possible to write a
> configure check for them?

The only problem I can see is that getcwd() is passed a fixed buffer size
which won't get reallocated if it's too small (the default is PATH_MAX
which should normally be enough), and there's no way to tell it to get as
much as it needs --- it always seems to use the size argument.  This could
be fixed by checking for an ERANGE and reallocating.  There's no record of
problems with real getcwd()s in the source, but I don't think it was used
at all till recently.  Currently configure doesn't even check for getcwd().
Does anyone know more?

-- 
Peter Stephenson <pws@ibmth.df.unipi.it>       Tel: +39 050 844536
WWW:  http://www.ifh.de/~pws/
Dipartimento di Fisica, Via Buonarroti 2, 56127 Pisa, Italy

Re: PATCH: 3.1.5-pws-12: getcwd

[Bart Schaefer]

On Mar 15,  1:32pm, Peter Stephenson wrote:
} Subject: Re: PATCH: 3.1.5-pws-12: getcwd
}
} "Andrej Borsenkow" wrote:
} > What is wrong with native getcwd? (Apart, that it may use native
} > malloc() to allocate buffer)? If there are any known problems, is it
} > possible to write a configure check for them?
} 
} The only problem I can see is that getcwd() is passed a fixed buffer size
} [...] Does anyone know more?

There's always the searchable archive ...

http://www.zsh.org/cgi-bin/mla/redirect?WORKERNUMBER=3480
http://www.zsh.org/cgi-bin/mla/redirect?WORKERNUMBER=3481

These articles ring a bell to the effect that zgetdir() is working around
some kind of problem with crossing of NFS mount points when walking up
the directory tree, affecting some but perhaps not all implementations
of get[c]wd().

-- 
Bart Schaefer                                 Brass Lantern Enterprises
http://www.well.com/user/barts              http://www.brasslantern.com