RE: Bash fails on BerkeleyDB configure

["Dave Korn" <dave dot korn at artimi dot com>]

On 26 November 2007 11:53, Howard Chu wrote:

[ Howard, it's not convention to Cc: the poster when replying on the Cygwin
list, but since I directed you here from the Gcc list where it is the SOP,
please let me know if you have a preference either way. ]

> if test $ac_cv_type_u_int8_t = yes; then
>    :
> else
> 
>      case "1" in
>      "$ac_cv_sizeof_unsigned_int")
>          u_int8_decl="typedef unsigned int u_int8_t;";;
>      "$ac_cv_sizeof_unsigned_char")
>          u_int8_decl="typedef unsigned char u_int8_t;";;
>      "$ac_cv_sizeof_unsigned_short")
>          u_int8_decl="typedef unsigned short u_int8_t;";;
>      "$ac_cv_sizeof_unsigned_long")
>          u_int8_decl="typedef unsigned long u_int8_t;";;
>      "$ac_cv_sizeof_unsigned_long_long")
>          u_int8_decl="typedef unsigned long long u_int8_t;";;
>      *)
>          if test "" != "notfatal"; then
>              { { echo "$as_me:$LINENO: error: No unsigned 1-byte integral
> type" >&5
> echo "$as_me: error: No unsigned 1-byte integral type" >&2;}
>     { (exit 1); exit 1; }; }
>          fi;;
>      esac
> fi
> 
> ac_cv_sizeof_unsigned_int is defined to 4 by prior tests. Likewise,
> ac_cv_sizeof_unsigned_char is 1
> ac_cv_sizeof_unsigned_short is 2
> ac_cv_sizeof_unsigned_long is 4
> ac_cv_sizeof_unsigned_long_long is 8
> 
> The test fails with "No unsigned 1-byte integral type". The script exits at
> that point. Getting past this (by explicitly setting variables to bypass the
> test) just runs into another failure as the script uses the same construct
> to set a bunch of other type defs.

  Thanks for sending the report.  The fact that it doesn't reproduce when you
paste it into a script by itself makes me suspicious that perhaps we're
getting line-ending trouble here: maybe ac_cv_sizeof_unsigned_char has
actually been defined to "1\r".  This can happen if there are any text-mode
mount points involved, or if the configure script has managed to get the wrong
kind of line-ends.

> Let me know if you need any more info.

  First thing I guess I need to ask is "You didn't by any chance use a
standard windows GUI app to unpack the BerkeleyDB tarball, did you"?  Things
like winzip et al., that aren't unix-aware, often convert line-endings from LF
to CRLF - which bash doesn't like.

  If it's not that simple, then can you please run "cygcheck -s -v -r >
cygcheck.out" on the affected machine, and post it to the list (as an
attachment please) with your next post?  That will give us an overview of the
setup and config of the cygwin installation on the affected machine.

    cheers,
      DaveK
-- 
Can't think of a witty .sigline today....


--
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/