Hi Derek,
I'm doing some testing of the tarball now: I'll have more later, but
a few quick comments/questions:
o Is it necessary to move the source modules up out of "src"? (I'd rather
not reorganize the CVS repository if not absolutely necessary, plus
I have a slight bias towards keeping the code in "src".)
o It looks like it builds the .o and binaries directly into the source
directories as opposed to a separate (obj, lib, bin) area. Is it
possible to continue to build into separate (obj, lib, bin) areas?
(In particular with the "arch" subdirectory to allow different
architectures to be built consecutively within the same source
build.)
o The two solaris builds at SLAC (sol8 and sol9) fail during "./configure"
with:
configure: error: Could not locate openssl/opensslv.h in /usr/include
It looks like this is in /usr/local/include at SLAC.
o It looks like the echo-ing of the compilation lines is on by default,
this was an option previously.
o For the Alice authorization package, we are happy to host this within
the xrootd CVS for convenience even if that means some extra configure
options. I think the important thing here is that those things are
packaged separately (e.g. in a separate rpm/tarball, etc.) by the
same build scripts.
Pete
On Mon, Aug 15, 2005 at 03:22:28PM +0200, Derek Feichtinger wrote:
> I have made the tarball with the autotools enabled build process available at
>
> http://people.web.psi.ch/feichtinger/xrootdnewbuild-20050728-0310c.tar.gz
> (this is the code of the current production version)
>
> No change to the sources was made (although some changes would have made
> possible a more elegant usage of some autotools features), but the
> distribution has the top directory layer removed, i.e. what you'll find is
> the former src/ directory.
>
> I provide some instructions in the INSTALL.* files. Please tell me, if you
> encounter any problems or if you would like any changes.
>
> There are some issues left:
> - what should be installed by 'make install'. Currently the binaries and some
> of the libraries get installed. But we should define the exact set of
> libraries and headers.
>
> - Where should the /etc and /utils end up. Should they be installed? The utils
> scripts would typically go to prefix/usr/bin. Where should we put the etc
> stuff? Maybe in prefix/etc/xrootd/?
>
> - The Alice authorization can be added to this in a separate module, which I
> will host presently (decided on not including them in the same tarball,
> because they require two other alice-specific libraries. This would require
> up to 5 additional options to the main configure command. I think it makes
> more sense to provide this as an optional download with its own configure
> script).
>
> - autotools also has some support for testing. It would be useful to add some
> standard tests.
>
> - maybe improving the 'configure --help' output
>
> Gerri and I have tested compilation of the tarball on a range of systems
> (Thanks for the help, Gerri!):
>
> CPU OS Compiler system's target string
> (as found by configure)
> ------------------------------------------------------------------------
> i686 RH 7.3 gcc-2.96 i686-pc-linux-gnu
> i686 (P4) SL CERN Release 3.0.5 gcc-3.2.3 i686-pc-linux-gnu
> i686 (AMD) RH EL3 gcc-3.2.3 i686-pc-linux-gnu
> i686 2.6.11-gentoo-r8 gcc-3.3.5 i686-pc-linux-gnu
> i686 debian SID (unstable) gcc-3.3.6 i686-pc-linux-gnu
> i686 slackware gcc-3.3.6 i686-pc-linux-gnu
> i686 RH 7.3 icc-8.0 i686-pc-linux-gnu
> IA-64 SL CERN Release 3.0.5 gcc-3.2.3 ia64-unknown-linux-gnu
> amd64 SL CERN Release 3 gcc-3.2.3 x86_64-unknown-linux-gnu
> powerpc darwin7.9.0 gcc-3.3 powerpc-apple-darwin7.9.0
> sparc Solaris 9 (SunOS) SunCC sparc-sun-solaris2.9
>
>
> However, it might be possible that building on other architectures requires
> additions to the configure.ac file.
> If you experience any problems, please send me the compile error and the
> contents of the config.log file.
>
> Cheers,
> Derek
>
> --
> Dr. Derek Feichtinger Tel: +41 22 767 10 07
> LCG/ARDA Group email: [log in to unmask]
> CERN
> CH-1211 Genève 23
>
-------------------------------------------------------------------------
Peter Elmer E-mail: [log in to unmask] Phone: +41 (22) 767-4644
Address: CERN Division PPE, Bat. 32 2C-14, CH-1211 Geneva 23, Switzerland
-------------------------------------------------------------------------
