 |
 |
Hi, from my point of view I'd be happy if it's possible. Another mini-issue. When committing something after having built the stuff, CVS complains about a missing GNUMakefile, because it has been renamed. It's a bit annoying to rename it back just in order to commit. Also, is there somewhere at slac a Solaris machine with updated autotools where I can build and test xrootd? Anyway I'll try to ask for an upgrade on the shires, we'll see. Fabrizio Peter Elmer wrote: > Hi Derek, > > Do you have any objection to simply maintaining an updated version > of ./configure in CVS? Although it is not quite as "from source" > as using the autotools to build it from scratch each time one checks > out from CVS, it would simplify things for most developers most of the > time. We just need to make sure it is updated when somebody makes changes > which require it to be updated. > > Pete > > On Tue, Oct 11, 2005 at 11:47:16AM +0200, Derek Feichtinger wrote: > >>Ciao, Fabrizio >> >>This is what I mentioned in my first migration mail to the list. You need >>reasonably new versions of the autotools (libtool should be very new to >>correctly produce files which work on Macs). >> >>------------------------------ >> >>>To use the new autotools build: >>>You should install automake > 1.6, autoconf-2.59 and libtool 1.5.18 on your >>>system. >> >>------------------------------- >> >>Best you get them as tarballs from the GNU site. They build very fast (first >>get Autoconf, then Automake and then libtool). The Solaris machines at SLAC >>seem to have really ancient versions which have been out of use for several >>years (early versions of Autoconf required a configure.in file. Now this has >>been renamed configure.ac to prevent confusion with the files on which >>configure acts, which are all named *.in) >> >>Cheers, >>Derek >> >> >>On Tuesday 11 October 2005 11.34, you wrote: >> >>>Hi, thank you. >>> >>>One more question. Is there anything special I have to do to work under >>>Solaris? >>>For instance, aclocal fails on shire02 at slac from a fresh checkout, it >>>says that it needs a configure.in file. From a Linux machine (noric02) >>>instead it works fine. >>> >>>Any hint? >>> >>> >>>Fabrizio >>> >>>Derek Feichtinger wrote: >>> >>>>Hi, Fabrizio >>>> >>>>Take a look at the Makefile.am in the XrdClient directory: >>>> >>>>If you want to recourse into subdirectories, you just add a >>>> >>>>SUBDIRS = dir1 dir2 dir3 >>>> >>>>statement. >>>> >>>>Build targets in automake are specified via directives like >>>> >>>>bin_PROGRAMS = myprogram1 myprogram2 >>>> >>>>The prefix (bin) defines where the file will be installed, the type >>>>(PROGRAM) defines what kind of targets should be built. >>>> >>>>There is a special prefix "check". This specifies targets that will never >>>>be installed and that get only built when a "make check" is given. This >>>>is specially designed for tests that are run after the initial build to >>>>check the package's sanity. >>>>The variable TESTS hold the name of the tests to be run. >>>> >>>>"Info Automake" should provide you with quite some information as well as >>>>the autobook (chapter 7.7): >>>>http://sources.redhat.com/autobook/autobook/autobook_toc.html >>>> >>>>I can gladly help you to get started if you have specific questions. >>>> >>>>Cheers, >>>>Derek >>>> >>>>On Tuesday 11 October 2005 10.35, Fabrizio Furano wrote: >>>> >>>>>Hi, >>>>> >>>>> just for the records, is there anything special I have to do if I want >>>>>to add new files/dirs to XrdClient with the new building scheme? >>>>> >>>>> If everything goes well, in the next period I'd like to add some >>>>>modules and a directory containing test programs. How do I do that? >>>>> >>>>>Fabrizio >>>>> >>>>>Derek Feichtinger wrote: >>>>> >>>>>>I forgot: >>>>>> >>>>>>to reduce the current compilation output, type >>>>>> >>>>>>make silent >>>>>> >>>>>>(just filters the compilation lines with sed, stderr still visible). At >>>>>>some point one could introduce a config.h file, which is used to pass >>>>>>all the compile switches determined by configure to the preprocessor, >>>>>>so the awfully long compile lines would get shorter. But this requires >>>>>>adding a include directive to all source files, so I did not want to do >>>>>>it now. >>>>>> >>>>>>To build the test files (e.g. for crypto and mon stuff) >>>>>> >>>>>>make check >>>>>> >>>>>>They will never be installed by make install (the check target is just >>>>>>for compiling and running a package's tests. At some point one could add >>>>>>some active tests to this target). >>>>>> >>>>>> >>>>>>make distcheck >>>>>> >>>>>>this will ensure, that a generated tarball can indeed compile. >>>>>> >>>>>>Cheers, >>>>>>Derek >> >>-- >>Dr. Derek Feichtinger Tel: +41 22 767 10 07 >>LCG/ARDA Group email: [log in to unmask] >>CERN http://people.web.psi.ch/feichtinger >>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 > -------------------------------------------------------------------------