You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@subversion.apache.org by Erik Huelsmann <e....@gmx.net> on 2004/03/20 16:14:25 UTC

Re: svn commit: r9168 - trunk/build/ac-macros

In standalone setups the HEAD of APR/APU_BRANCH_0_9 installs into
/usr/local/apr by default. I committed this change in order for it to be ported to the
1.0.x branch for better user experience. (I will nominate it soon.)

Re: svn commit: r9168 - trunk/build/ac-macros

Posted by Erik Huelsmann <e....@gmx.net>.
> --On Monday, March 29, 2004 10:06 PM +0200 Erik Huelsmann 
> <e....@gmx.net> wrote:
> 
> > Forgive my ignorance, but how is this confusion different if apr is
> > compiling to /usr/lib from compiling to /usr/local/apr/lib?
> 
> It isn't.  But, libtool has some explicit checks to help reduce the risk
> in the /usr/lib and /usr/local/lib case - mainly by not adding those
explicit
> link paths - which skirts the issue somewhat for some people.

done. r9125.

bye.

Erik.

-- 
+++ NEU bei GMX und erstmalig in Deutschland: TÜV-geprüfter Virenschutz +++
100% Virenerkennung nach Wildlist. Infos: http://www.gmx.net/virenschutz


---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: svn commit: r9168 - trunk/build/ac-macros

Posted by Justin Erenkrantz <ju...@erenkrantz.com>.
--On Monday, March 29, 2004 10:06 PM +0200 Erik Huelsmann 
<e....@gmx.net> wrote:

> Forgive my ignorance, but how is this confusion different if apr is
> compiling to /usr/lib from compiling to /usr/local/apr/lib?

It isn't.  But, libtool has some explicit checks to help reduce the risk in 
the /usr/lib and /usr/local/lib case - mainly by not adding those explicit 
link paths - which skirts the issue somewhat for some people.

> I noticed this, since *we* may be distributing apr in our tarball, but I
> found some source rpms which did not include the neon or apr{,-util}
> libraries. Presumably because those are distributed in other srpms.
> Although we can't take just any packager into account, it is not just any
> user which is confronted with this "problem".

It's the packager's responsibility not ours.  Using an installed APR should 
be done only with explicit knowledge - the implicit install detection 
option is really a loaded gun.  As such, it's our last resort - with our 
normal packaging not allowing it at all.

When we had that enabled, you should have seen all the bug reports and 
failures we had.  I'd recommend looking in the list archives...  -- justin

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: svn commit: r9168 - trunk/build/ac-macros

Posted by Erik Huelsmann <e....@gmx.net>.
> > In standalone setups the HEAD of APR/APU_BRANCH_0_9 installs into
> > /usr/local/apr by default. I committed this change in order for it to be
> > ported to the 1.0.x branch for better user experience. (I will nominate
> > it soon.)
> >
> > From discussion on this list, I concluded that these two files are
> copies
> > from the apr(-util) provided ones in the apache CVS repository. Possibly
> > one of the APR developers on the list want to take this commit there
> too?
> 
> No.  This should probably be reverted.  Kevin had that there initially,
> but 
> the problem is that libtool can be confused on a previous install when it 
> is compiling to that same location.  This is why there is the 
> 'implicit-install-check'. 

Forgive my ignorance, but how is this confusion different if apr is
compiling to /usr/lib from compiling to /usr/local/apr/lib?

I noticed this, since *we* may be distributing apr in our tarball, but I
found some source rpms which did not include the neon or apr{,-util} libraries.
Presumably because those are distributed in other srpms. Although we can't
take just any packager into account, it is not just any user which is
confronted with this "problem".

bye,

Erik.


-- 
+++ NEU bei GMX und erstmalig in Deutschland: TÜV-geprüfter Virenschutz +++
100% Virenerkennung nach Wildlist. Infos: http://www.gmx.net/virenschutz


---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: svn commit: r9168 - trunk/build/ac-macros

Posted by Justin Erenkrantz <ju...@erenkrantz.com>.
--On Saturday, March 20, 2004 5:14 PM +0100 Erik Huelsmann 
<e....@gmx.net> wrote:

> In standalone setups the HEAD of APR/APU_BRANCH_0_9 installs into
> /usr/local/apr by default. I committed this change in order for it to be
> ported to the 1.0.x branch for better user experience. (I will nominate
> it soon.)
>
> From discussion on this list, I concluded that these two files are copies
> from the apr(-util) provided ones in the apache CVS repository. Possibly
> one of the APR developers on the list want to take this commit there too?

No.  This should probably be reverted.  Kevin had that there initially, but 
the problem is that libtool can be confused on a previous install when it 
is compiling to that same location.  This is why there is the 
'implicit-install-check'.  Furthermore, this'll only get used when there is 
no apr and apr-util subdirs - which we always include in our release 
tarballs.  So, the value to our naïve end users is minimal.

The user should just explicitly pass the --with-apr and --with-apr-util 
flags - the risk of corruption is way too high here.  -- justin

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org