RE: autoconf issues [often when upgrading from 2.13->2.53a]

RE: autoconf issues [often when upgrading from 2.13->2.53a]

[Harig, Mark A.]

I'm using Win2K.  

I'm pretty reluctant to point the finger at 'sed', but unfortunately I
don't have any more clues at this time.  Can you add 'set -xv' to the
generated configure script at the place in the code where 'sed' is
called, and post the result?  That, along with the input stream, could
help to debug that problem.  Also, it would help to know what
configure.in statements are generating the corresponding code in the
generated configure script, if you're able to identify those statements.

As usual, three basics are needed before the problem can be debugged:

  1. The command(s) that are "causing" the problem.
  2. The results you expected.
  3. The results you got.

> -----Original Message-----
> From: Nicholas Wourms [mailto:nwourms@yahoo.com]
> Sent: Monday, July 22, 2002 6:55 PM
> To: cygwin@cygwin.com
> Subject: RE: autoconf issues [often when upgrading from 2.13->2.53a]
> 
> 
> 
> --- "Harig, Mark A." <maharig@idirect.net> wrote:
> > Have you tried downgrading to the stable tools (2.13) to see if the
> > problem is still occurring, or is that not an option (i.e, the
> > configure.in requires 2.53)?
> > I had a problem recently where it appeared that sed was running in
> > an
> > infinite loop.  When I downgraded to the stable tools, the problem
> > disappeared.
> 
> Unfortunately I have to use 2.53, otherwise libtool will not build
> the dll's properly.  You hit the nail right on the head, there is
> something awefully wierd going on, and I think sed is at the center
> of it.  Why just the other day I was trying to use texi2pdf, and
> wouldn't you know it sed was stuck in an infinite loop.  What OS are
> you running?  My initial opinion is that the sed is experiencing
> bit-rot.  It hasn't been updated since 2000 (the code itself since
> '98), you know.  The problem is that this is so random, there's no
> easy way to tell for sure.  I just have a hunch at this point...
> 
> Cheers,
> Nicholas
> 
> __________________________________________________
> Do You Yahoo!?
> Yahoo! Health - Feel better, live better
> http://health.yahoo.com
> 
> --
> Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
> Bug reporting:         http://cygwin.com/bugs.html
> Documentation:         http://cygwin.com/docs.html
> FAQ:                   http://cygwin.com/faq/
> 
> 

--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/

RE: autoconf issues [often when upgrading from 2.13->2.53a]

[Nicholas Wourms]

Mark,

Are you aware of how many instances of sed autoconf runs, let alone
libtool?  I'll post the statement for the original issue this thread
describes, but as for anything else, that will be more difficult. 
Unfortunately TeX is very convoluted in it's own way, so trying to
figure out what pdfTeX is doing with sed is like searching for a
needle in the haystack.  I'll see what I can do.

Cheers,
Nicholas
> I'm using Win2K.  
> 
> I'm pretty reluctant to point the finger at 'sed', but
> unfortunately I
> don't have any more clues at this time.  Can you add 'set -xv' to
> the
> generated configure script at the place in the code where 'sed' is
> called, and post the result?  That, along with the input stream,
> could
> help to debug that problem.  Also, it would help to know what
> configure.in statements are generating the corresponding code in
> the
> generated configure script, if you're able to identify those
> statements.
> 
> As usual, three basics are needed before the problem can be
> debugged:
> 
>   1. The command(s) that are "causing" the problem.
>   2. The results you expected.
>   3. The results you got.
> 
> > -----Original Message-----
> > From: Nicholas Wourms [mailto:nwourms@yahoo.com]
> > Sent: Monday, July 22, 2002 6:55 PM
> > To: cygwin@cygwin.com
> > Subject: RE: autoconf issues [often when upgrading from
> 2.13->2.53a]
> > 
> > 
> > 
> > --- "Harig, Mark A." <maharig@idirect.net> wrote:
> > > Have you tried downgrading to the stable tools (2.13) to see if
> the
> > > problem is still occurring, or is that not an option (i.e, the
> > > configure.in requires 2.53)?
> > > I had a problem recently where it appeared that sed was running
> in
> > > an
> > > infinite loop.  When I downgraded to the stable tools, the
> problem
> > > disappeared.
> > 
> > Unfortunately I have to use 2.53, otherwise libtool will not
> build
> > the dll's properly.  You hit the nail right on the head, there is
> > something awefully wierd going on, and I think sed is at the
> center
> > of it.  Why just the other day I was trying to use texi2pdf, and
> > wouldn't you know it sed was stuck in an infinite loop.  What OS
> are
> > you running?  My initial opinion is that the sed is experiencing
> > bit-rot.  It hasn't been updated since 2000 (the code itself
> since
> > '98), you know.  The problem is that this is so random, there's
> no
> > easy way to tell for sure.  I just have a hunch at this point...


__________________________________________________
Do You Yahoo!?
Yahoo! Health - Feel better, live better
http://health.yahoo.com

--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/

RE: autoconf issues [often when upgrading from 2.13->2.53a]

[Harig, Mark A.]

>  
> > Looking at the log of the configure session that you included, it's
> > apparent that your version of 'expr' is returning the wrong result.
> > Here's the command that the configure script runs:
> > 
> > expr xposixmutexes :
> >
> '.*[^-_abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ012
> 3456789]'
> > && echo hello
> > 
> > (I changed the 'echo' portion for simplicity.)
> > 
> > This invocation of 'expr' should return 0 (zero) and the second
> > half of
> > the '&&' expression should not be evaluated, but on your system it
> > is.
> 
> Hmm, I see...
>  
> > You might try running the first half of the '&&' expression without
> > the
> > second half to see what position in the 'xposixmutexes' string
> > 'expr'
> > thinks it is finding an invalid character.
> 
> Curious, this is the result on the Win98SE machine:
> $ expr xposixmutexes :
> '.*[^-_abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ012
> 3456789]'
> 
> 0
> 
> I take it this isn't what should have happened if something was
> wrong?  However, I immediately try the configure script and I get
> that error...
>

My guess is that when the 'configure' script is running, it is finding
a different version of 'expr'.  Is it possible that you have more than
one version installed on your machine?

At your shell prompt, enter the command:

   $ type expr

Next, edit your 'configure' script and find the section that evaluates
the '--enable' option, i.e., 

  -enable-* | --enable-*)
    ac_feature=`expr "x$ac_option" : 'x-*enable-\([^=]*\)'`
    # Reject names that are not valid shell variable names.
    expr "x$ac_feature" : ".*[^-_$as_cr_alnum]" >/dev/null &&
      { echo "$as_me: error: invalid feature name: $ac_feature" >&2
   { (exit 1); exit 1; }; }
    ac_feature=`echo $ac_feature | sed 's/-/_/g'`
    case $ac_option in
      *=*) ac_optarg=`echo "$ac_optarg" | sed "s/'/'\\\\\\\\''/g"`;;
      *) ac_optarg=yes ;;
    esac
    eval "enable_$ac_feature='$ac_optarg'" ;;

Insert the line 'type expr' before the line:

>    expr "x$ac_feature" : ".*[^-_$as_cr_alnum]" >/dev/null &&

Save the change and run the 'configure' script again.  It should fail
again, of course, but it should also tell you where it thinks it is
finding the 'expr' program.

By the way, your instructions for generating 'configure' using your
'db-3.1.17-1.sh' script did not mention that you had overridden the
settings of $CC and $CXX to use 'gcc-2' and 'g++-2'.

---

--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/

RE: autoconf issues [often when upgrading from 2.13->2.53a]

[Nicholas Wourms]

--- "Harig, Mark A." <maharig@idirect.net> wrote:
> First some facts about what I have:
> 
> 1. I have access to Win2K, but not 98/ME/etc.

I'm running both WinME and a Win98SE machines.
 
> 2. cygwin, version 1.3.12-2

WinME: v1.3.13(CVS)
Win98SE: v1.3.12-2
 
> 3. $ bash --version
>    GNU bash, version 2.05a.0(3)-release (i686-pc-cygwin)
>    Copyright 2001 Free Software Foundation, Inc.

Both are running the latest bash from the mirrors:
GNU bash, version 2.05b.0(1)-release
 
> 4. $ expr --version
>    expr (GNU sh-utils) 2.0

Both are running expr (GNU sh-utils) 2.0
 
> Looking at the log of the configure session that you included, it's
> apparent that your version of 'expr' is returning the wrong result.
> Here's the command that the configure script runs:
> 
> expr xposixmutexes :
>
'.*[^-_abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789]'
> && echo hello
> 
> (I changed the 'echo' portion for simplicity.)
> 
> This invocation of 'expr' should return 0 (zero) and the second
> half of
> the '&&' expression should not be evaluated, but on your system it
> is.

Hmm, I see...
 
> You might try running the first half of the '&&' expression without
> the
> second half to see what position in the 'xposixmutexes' string
> 'expr'
> thinks it is finding an invalid character.

Curious, this is the result on the Win98SE machine:
$ expr xposixmutexes :
'.*[^-_abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789]'

0

I take it this isn't what should have happened if something was
wrong?  However, I immediately try the configure script and I get
that error...
 
> Are you running the latest stable version of 'expr' that Cygwin has
> available?  It's included in sh-utils 2.0-2, which is in the 'Base'
> category when setup.exe is run.
> 

Yes, I'm always running the latest :-).  As a side note, eventhough
I'm regex-illiterate, I still know what I'm doing for the most part
;-).

Cheers,
Nicholas

__________________________________________________
Do You Yahoo!?
Yahoo! Health - Feel better, live better
http://health.yahoo.com

--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/

RE: autoconf issues [often when upgrading from 2.13->2.53a]

[Harig, Mark A.]

One other thing: You might try using the alternative syntax for the
'expr' command, that is, instead of:

   expr STRING : REGEXP

try

   expr match STRING REGEXP

Specifically,

   $ expr match xposixmutexes
'.*[^-_abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789]'

should return 0 (zero).

> -----Original Message-----
> From: Harig, Mark A. 
> Sent: Tuesday, July 23, 2002 7:09 PM
> To: Nicholas Wourms; cygwin@cygwin.com
> Subject: RE: autoconf issues [often when upgrading from 2.13->2.53a]
> 
> 
> First some facts about what I have:
> 
> 1. I have access to Win2K, but not 98/ME/etc.
> 
> 2. cygwin, version 1.3.12-2
> 
> 3. $ bash --version
>    GNU bash, version 2.05a.0(3)-release (i686-pc-cygwin)
>    Copyright 2001 Free Software Foundation, Inc.
> 
> 4. $ expr --version
>    expr (GNU sh-utils) 2.0
> 
> Looking at the log of the configure session that you included, it's
> apparent that your version of 'expr' is returning the wrong result.
> Here's the command that the configure script runs:
> 
> expr xposixmutexes :
> '.*[^-_abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ012
> 3456789]'
> && echo hello
> 
> (I changed the 'echo' portion for simplicity.)
> 
> This invocation of 'expr' should return 0 (zero) and the 
> second half of
> the '&&' expression should not be evaluated, but on your system it is.
> 
> You might try running the first half of the '&&' expression 
> without the
> second half to see what position in the 'xposixmutexes' string 'expr'
> thinks it is finding an invalid character.
> 
> Are you running the latest stable version of 'expr' that Cygwin has
> available?  It's included in sh-utils 2.0-2, which is in the 'Base'
> category when setup.exe is run.
> 
> > 
> > I have attached a log of the configure session with set -xv.  Don't
> > worry, it bombs out early, so it isn't that big.  I'll do the same
> > procedure on win2k, if you want.  The offending checkpoint is:
> > 
> >   -enable-* | --enable-*)
> >     ac_feature=`expr "x$ac_option" : 'x-*enable-\([^=]*\)'`
> >     # Reject names that are not valid shell variable names.
> >     expr "x$ac_feature" : ".*[^-_$as_cr_alnum]" >/dev/null &&
> >       { echo "$as_me: error: invalid feature name: $ac_feature" >&2
> >    { (exit 1); exit 1; }; }   
> >     ac_feature=`echo $ac_feature | sed 's/-/_/g'`
> >     case $ac_option in
> >       *=*) ac_optarg=`echo "$ac_optarg" | sed "s/'/'\\\\\\\\''/g"`;;
> >       *) ac_optarg=yes ;;
> >     esac
> >     eval "enable_$ac_feature='$ac_optarg'" ;;
> > 
> > The options I'm passing are:
> > 
> > CFLAGS=-O2 \
> > LDFLAGS= \
> > CC=gcc-2 \
> > CXX=g++-2 \
> > ../dist/configure --target=i686-pc-cygwin --prefix=/usr
> > --exec-prefix=/usr --sysconfdir=/etc --bindir=/usr/bin
> > --libdir=/usr/lib --includedir=/usr/include/db3.1
> > --datadir=/usr/share '--libexecdir=${sbindir}' --localstatedir=/var
> > '--program-transform-name=s,^db_,db3.1_,' --enable-posixmutexes
> > --enable-dynamic --enable-cxx --enable-compat185 --disable-test
> > 
> 
> --
> Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
> Bug reporting:         http://cygwin.com/bugs.html
> Documentation:         http://cygwin.com/docs.html
> FAQ:                   http://cygwin.com/faq/
> 
> 

--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/

RE: autoconf issues [often when upgrading from 2.13->2.53a]

[Harig, Mark A.]

First some facts about what I have:

1. I have access to Win2K, but not 98/ME/etc.

2. cygwin, version 1.3.12-2

3. $ bash --version
   GNU bash, version 2.05a.0(3)-release (i686-pc-cygwin)
   Copyright 2001 Free Software Foundation, Inc.

4. $ expr --version
   expr (GNU sh-utils) 2.0

Looking at the log of the configure session that you included, it's
apparent that your version of 'expr' is returning the wrong result.
Here's the command that the configure script runs:

expr xposixmutexes :
'.*[^-_abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789]'
&& echo hello

(I changed the 'echo' portion for simplicity.)

This invocation of 'expr' should return 0 (zero) and the second half of
the '&&' expression should not be evaluated, but on your system it is.

You might try running the first half of the '&&' expression without the
second half to see what position in the 'xposixmutexes' string 'expr'
thinks it is finding an invalid character.

Are you running the latest stable version of 'expr' that Cygwin has
available?  It's included in sh-utils 2.0-2, which is in the 'Base'
category when setup.exe is run.

> 
> I have attached a log of the configure session with set -xv.  Don't
> worry, it bombs out early, so it isn't that big.  I'll do the same
> procedure on win2k, if you want.  The offending checkpoint is:
> 
>   -enable-* | --enable-*)
>     ac_feature=`expr "x$ac_option" : 'x-*enable-\([^=]*\)'`
>     # Reject names that are not valid shell variable names.
>     expr "x$ac_feature" : ".*[^-_$as_cr_alnum]" >/dev/null &&
>       { echo "$as_me: error: invalid feature name: $ac_feature" >&2
>    { (exit 1); exit 1; }; }   
>     ac_feature=`echo $ac_feature | sed 's/-/_/g'`
>     case $ac_option in
>       *=*) ac_optarg=`echo "$ac_optarg" | sed "s/'/'\\\\\\\\''/g"`;;
>       *) ac_optarg=yes ;;
>     esac
>     eval "enable_$ac_feature='$ac_optarg'" ;;
> 
> The options I'm passing are:
> 
> CFLAGS=-O2 \
> LDFLAGS= \
> CC=gcc-2 \
> CXX=g++-2 \
> ../dist/configure --target=i686-pc-cygwin --prefix=/usr
> --exec-prefix=/usr --sysconfdir=/etc --bindir=/usr/bin
> --libdir=/usr/lib --includedir=/usr/include/db3.1
> --datadir=/usr/share '--libexecdir=${sbindir}' --localstatedir=/var
> '--program-transform-name=s,^db_,db3.1_,' --enable-posixmutexes
> --enable-dynamic --enable-cxx --enable-compat185 --disable-test
> 

--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/

RE: autoconf issues [often when upgrading from 2.13->2.53a]

[Nicholas Wourms]

[-- Attachment #1: Type: text/plain, Size: 2596 bytes --]

--- "Harig, Mark A." <maharig@idirect.net> wrote:
> Yes, autoconf uses sed many, many times.  What is often possible is
> to
> narrow down which one by searching in 'configure' for the text
> message
> that 'configure' generates at the time that the script hangs, e.g.,
> "Checking for GNU c compiler..."  Sometimes you may need to do this
> several times.  The better approach is to have a firm understanding
> of
> the layers of abstraction that autoconf builds on top of m4 and
> automake
> builds on top of autoconf, etc. but we're trying the
> "brute-force-examine-the-assembly-language" approach.  'config.log'
> is
> sometimes helpful in providing clues about what might be causing
> the
> problem.

Mark,

I have attached a log of the configure session with set -xv.  Don't
worry, it bombs out early, so it isn't that big.  I'll do the same
procedure on win2k, if you want.  The offending checkpoint is:

  -enable-* | --enable-*)
    ac_feature=`expr "x$ac_option" : 'x-*enable-\([^=]*\)'`
    # Reject names that are not valid shell variable names.
    expr "x$ac_feature" : ".*[^-_$as_cr_alnum]" >/dev/null &&
      { echo "$as_me: error: invalid feature name: $ac_feature" >&2
   { (exit 1); exit 1; }; }   
    ac_feature=`echo $ac_feature | sed 's/-/_/g'`
    case $ac_option in
      *=*) ac_optarg=`echo "$ac_optarg" | sed "s/'/'\\\\\\\\''/g"`;;
      *) ac_optarg=yes ;;
    esac
    eval "enable_$ac_feature='$ac_optarg'" ;;

The options I'm passing are:

CFLAGS=-O2 \
LDFLAGS= \
CC=gcc-2 \
CXX=g++-2 \
../dist/configure --target=i686-pc-cygwin --prefix=/usr
--exec-prefix=/usr --sysconfdir=/etc --bindir=/usr/bin
--libdir=/usr/lib --includedir=/usr/include/db3.1
--datadir=/usr/share '--libexecdir=${sbindir}' --localstatedir=/var
'--program-transform-name=s,^db_,db3.1_,' --enable-posixmutexes
--enable-dynamic --enable-cxx --enable-compat185 --disable-test

If you (or anyone else) are interested in seeing what it does on your
system, you can grab the self-contained build system at:

http://today.clemson.edu/temp/db-3.1.17-1-src.tar.bz2

Untar to a temporary dir and run the buildscript like so:
./db-3.1.17-1.sh prep

Then edit the configure script [db-3.1.17/dist/configure] adding your
-xv flags to line 1 and line 155 [exec "$CONFIG_SHELL" "-xv"]:

Then run:
./db-3.1.17-1.sh conf

It should run fine on win2k, but winme/9x it doesn't.  I'm going to
try and examine it some more, but I thought I'd bring you in on it.

Cheers,
Nicholas


__________________________________________________
Do You Yahoo!?
Yahoo! Health - Feel better, live better
http://health.yahoo.com

[-- Attachment #2: configure.log.bz2 --]
[-- Type: application/octet-stream, Size: 9718 bytes --]

[-- Attachment #3: Type: text/plain, Size: 214 bytes --]

--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/

RE: autoconf issues [often when upgrading from 2.13->2.53a]

[Harig, Mark A.]

Yes, autoconf uses sed many, many times.  What is often possible is to
narrow down which one by searching in 'configure' for the text message
that 'configure' generates at the time that the script hangs, e.g.,
"Checking for GNU c compiler..."  Sometimes you may need to do this
several times.  The better approach is to have a firm understanding of
the layers of abstraction that autoconf builds on top of m4 and automake
builds on top of autoconf, etc. but we're trying the
"brute-force-examine-the-assembly-language" approach.  'config.log' is
sometimes helpful in providing clues about what might be causing the
problem.

> -----Original Message-----
> From: Nicholas Wourms [mailto:nwourms@yahoo.com]
> Sent: Monday, July 22, 2002 8:58 PM
> To: cygwin@cygwin.com
> Subject: RE: autoconf issues [often when upgrading from 2.13->2.53a]
> 
> 
> Mark,
> 
> Are you aware of how many instances of sed autoconf runs, let alone
> libtool?  I'll post the statement for the original issue this thread
> describes, but as for anything else, that will be more difficult. 
> Unfortunately TeX is very convoluted in it's own way, so trying to
> figure out what pdfTeX is doing with sed is like searching for a
> needle in the haystack.  I'll see what I can do.
> 
> Cheers,
> Nicholas
> > I'm using Win2K.  
> > 
> > I'm pretty reluctant to point the finger at 'sed', but
> > unfortunately I
> > don't have any more clues at this time.  Can you add 'set -xv' to
> > the
> > generated configure script at the place in the code where 'sed' is
> > called, and post the result?  That, along with the input stream,
> > could
> > help to debug that problem.  Also, it would help to know what
> > configure.in statements are generating the corresponding code in
> > the
> > generated configure script, if you're able to identify those
> > statements.
> > 
> > As usual, three basics are needed before the problem can be
> > debugged:
> > 
> >   1. The command(s) that are "causing" the problem.
> >   2. The results you expected.
> >   3. The results you got.
> > 
> > > -----Original Message-----
> > > From: Nicholas Wourms [mailto:nwourms@yahoo.com]
> > > Sent: Monday, July 22, 2002 6:55 PM
> > > To: cygwin@cygwin.com
> > > Subject: RE: autoconf issues [often when upgrading from
> > 2.13->2.53a]
> > > 
> > > 
> > > 
> > > --- "Harig, Mark A." <maharig@idirect.net> wrote:
> > > > Have you tried downgrading to the stable tools (2.13) to see if
> > the
> > > > problem is still occurring, or is that not an option (i.e, the
> > > > configure.in requires 2.53)?
> > > > I had a problem recently where it appeared that sed was running
> > in
> > > > an
> > > > infinite loop.  When I downgraded to the stable tools, the
> > problem
> > > > disappeared.
> > > 
> > > Unfortunately I have to use 2.53, otherwise libtool will not
> > build
> > > the dll's properly.  You hit the nail right on the head, there is
> > > something awefully wierd going on, and I think sed is at the
> > center
> > > of it.  Why just the other day I was trying to use texi2pdf, and
> > > wouldn't you know it sed was stuck in an infinite loop.  What OS
> > are
> > > you running?  My initial opinion is that the sed is experiencing
> > > bit-rot.  It hasn't been updated since 2000 (the code itself
> > since
> > > '98), you know.  The problem is that this is so random, there's
> > no
> > > easy way to tell for sure.  I just have a hunch at this point...
> 
> 
> __________________________________________________
> Do You Yahoo!?
> Yahoo! Health - Feel better, live better
> http://health.yahoo.com
> 
> --
> Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
> Bug reporting:         http://cygwin.com/bugs.html
> Documentation:         http://cygwin.com/docs.html
> FAQ:                   http://cygwin.com/faq/
> 
> 

--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/

RE: autoconf issues [often when upgrading from 2.13->2.53a]

[Nicholas Wourms]

--- "Harig, Mark A." <maharig@idirect.net> wrote:
> Have you tried downgrading to the stable tools (2.13) to see if the
> problem is still occurring, or is that not an option (i.e, the
> configure.in requires 2.53)?
> I had a problem recently where it appeared that sed was running in
> an
> infinite loop.  When I downgraded to the stable tools, the problem
> disappeared.

Unfortunately I have to use 2.53, otherwise libtool will not build
the dll's properly.  You hit the nail right on the head, there is
something awefully wierd going on, and I think sed is at the center
of it.  Why just the other day I was trying to use texi2pdf, and
wouldn't you know it sed was stuck in an infinite loop.  What OS are
you running?  My initial opinion is that the sed is experiencing
bit-rot.  It hasn't been updated since 2000 (the code itself since
'98), you know.  The problem is that this is so random, there's no
easy way to tell for sure.  I just have a hunch at this point...

Cheers,
Nicholas

__________________________________________________
Do You Yahoo!?
Yahoo! Health - Feel better, live better
http://health.yahoo.com

--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/

RE: autoconf issues [often when upgrading from 2.13->2.53a]

[Harig, Mark A.]

Have you tried downgrading to the stable tools (2.13) to see if the
problem is still occurring, or is that not an option (i.e, the
configure.in requires 2.53)?
I had a problem recently where it appeared that sed was running in an
infinite loop.  When I downgraded to the stable tools, the problem
disappeared.

> -----Original Message-----
> From: Nicholas Wourms [mailto:nwourms@yahoo.com]
> Sent: Saturday, July 20, 2002 10:48 AM
> To: cygwin@cygwin.com
> Subject: autoconf issues [often when upgrading from 2.13->2.53a]
> 
> 
> Hi,
> 
> I've been experiencing some strange problems with the autotools as of
> late.  First, let me tell you what I have installed.  I have 
> updated to
> the latest version of all the autotools and wrapper scripts.  
> I've been
> working on a few packages which required re-autotoolizing to 
> the latest
> versions to support shared dll's.  The problem is that 
> lately, many of the
> packages get petulant when I try to pass options to them.  
> 
> For example:
> ./configure --prefix=/usr ... --enable-shared
> 
> will return:
> shared: invalid feature name
> 
> I've looked on google and I've check the mailing lists, but I 
> can't seem
> to get a definitive answer on what might cause this to 
> happen.  I tried to
> follow the autotools script, but it isn't the easiest thing 
> to unravel. 
> To get around this, I have to manually edit the configure 
> file and remove
> the 3 lines that do the validity test.  I don't think this is 
> the proper
> solution, in the long run.  So, my hope is that someone has 
> experienced
> this lately and can clue me in on a good way isolate what is 
> causing this.
>  I might note that this behaviour started happening after I 
> upgraded to
> automake-1.6.2, but I can't say for sure that this is the cause of my
> woes.  I wanted to atleast see if anyone else is aware of 
> this on cygwin
> before I go posting to the autoconf list.  Maybe I'm not thinking too
> clearly and have been working too hard on one of my packages, 
> but this is
> driving me crazy.  Any helpful pointers or references to some
> documentation I might have missed would be greatly appreciated.
> 
> Cheers,
> Nicholas
> 
> __________________________________________________
> Do You Yahoo!?
> Yahoo! Health - Feel better, live better
> http://health.yahoo.com
> 
> --
> Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
> Bug reporting:         http://cygwin.com/bugs.html
> Documentation:         http://cygwin.com/docs.html
> FAQ:                   http://cygwin.com/faq/
> 
> 

--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/

autoconf issues [often when upgrading from 2.13->2.53a]

[Nicholas Wourms]

Hi,

I've been experiencing some strange problems with the autotools as of
late.  First, let me tell you what I have installed.  I have updated to
the latest version of all the autotools and wrapper scripts.  I've been
working on a few packages which required re-autotoolizing to the latest
versions to support shared dll's.  The problem is that lately, many of the
packages get petulant when I try to pass options to them.  

For example:
./configure --prefix=/usr ... --enable-shared

will return:
shared: invalid feature name

I've looked on google and I've check the mailing lists, but I can't seem
to get a definitive answer on what might cause this to happen.  I tried to
follow the autotools script, but it isn't the easiest thing to unravel. 
To get around this, I have to manually edit the configure file and remove
the 3 lines that do the validity test.  I don't think this is the proper
solution, in the long run.  So, my hope is that someone has experienced
this lately and can clue me in on a good way isolate what is causing this.
 I might note that this behaviour started happening after I upgraded to
automake-1.6.2, but I can't say for sure that this is the cause of my
woes.  I wanted to atleast see if anyone else is aware of this on cygwin
before I go posting to the autoconf list.  Maybe I'm not thinking too
clearly and have been working too hard on one of my packages, but this is
driving me crazy.  Any helpful pointers or references to some
documentation I might have missed would be greatly appreciated.

Cheers,
Nicholas

__________________________________________________
Do You Yahoo!?
Yahoo! Health - Feel better, live better
http://health.yahoo.com

--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/