Re: [2.63] - problem parsing generated configure or config.status

[Unknown]

On Tue, 23 Jun 2009, Ralf Wildenhues wrote:

> I see this construct in the generated config.status file, but I can't
> see anything wrong with it (the final ;; ends a case statement).  I'm
> afraid you might have to debug it yourself, as I don't have any systems
> currently running with bash 2.05b installed any more, and I've confirmed
> your observation that it doesn't hit with a self-compiled 2.05 on
> GNU/Linux.  This shell is 7 years old and Cygwin is easily updated, why
> not move to a newer version?
Well. I have my reasons. To make long story short, cygwin-1.3.2 is
small, fast and stable. I tried to upgrade once to cygwin-1.5. I lost 
peformance and even some apps wouldnt work at all under it (aka screen).

> That said, if you can analyze the issue and we find an easy workaround,
> we can still consider it.
Okey, seems I found sth.. As I said.. I cannot compile autoconf on my 
platform so I was started by tinkering configure script itself.
Here are results:


% diff -ub configure configure.my
--- configure   Mon Jun 22 08:24:26 2009
+++ configure.my        Wed Jun 24 12:43:37 2009
@@ -7945,7 +7945,7 @@
 if test -n "$CONFIG_FILES"; then


'ac_cr='
+ac_cr=''
 ac_cs_awk_cr=`$AWK 'BEGIN { print "a\rb" }' </dev/null 2>/dev/null`
 if test "$ac_cs_awk_cr" = "a${ac_cr}b"; then
   ac_cs_awk_cr='\\r'


Now, dont get fulled by that strange:
'ac_cr='
because in VIM it looks like this:
ac_cr='^M'
Seems its a carriage return code that breaks cygwin bash.
When I simply removed it (as in diff) everything went smooth
and I managed to compile git-1.6.3.3 on my platform.

Hope now, this will point you in right direction.

Regards,
Borg