Re: [OPEN-ILS-GENERAL] Problem with yaz, debian-squeeze and Net::Z3950-SimpleServer
On Thu, Oct 6, 2011 at 9:11 PM, Dan Scott d...@coffeecode.net wrote: snip Net-SimpleServer-Z3950 1.12 works happily with Debian Squeeze. Don't have the command at my fingertips but something like the following from the CPAN shell should work: cpan install M/MI/MIRK/Net-SimpleServer-Z3950-1.12 Correction, now that I have a dev system ready at hand - the command should be: cpan install M/MI/MIRK/Net-Z3950-SimpleServer-1.12.tar.gz
Re: [OPEN-ILS-GENERAL] Problem with yaz, debian-squeeze and Net::Z3950-SimpleServer
On Thu, 2011-10-06 at 21:11 -0400, Dan Scott wrote: You have to install a previous version of Net::Z3950::SimpleServer from CPAN. There were hopes that the maintainers would realize that it would be nice if their package would install on the most recent stable Debian release and make the dependency on the most recent source version of yaz optional, but that seems not to be in the cards. Net-SimpleServer-Z3950 1.12 works happily with Debian Squeeze. Don't have the command at my fingertips but something like the following from the CPAN shell should work: cpan install M/MI/MIRK/Net-SimpleServer-Z3950-1.12 Ok, that is solved. Had to apt-get install yaz first but that might have been me installing and reinstalling things trying in vain to make it work. This is a relatively recent development (well - middle of August) and as the package is only required for providing a Z39.50 server it hasn't been on the front of people's minds; they generally don't run into it when testing out the basic functions of an installed system. We should be able to tweak Makefile.install accordingly - either to install the latest yaz from source (sigh - hopefully _that_ won't introduce problems to more core functions like z39.50 import...) or to install Net::SimpleServer::Z3950 1.12 specifically (I'd lean towards that because it has less impact on the system as a whole). Well I am running 1.6.0.8 in production right now and and the idea is to get this pair of machines up and try a test import of our data with an idea to migrate after letting the staff beat on it a week or so. So I was watching the install logs carefully for errors since the plan is for this install to go into production. I am working on the assumption that I can stuff a copy of our data in, run each of the schema update scripts and be ready to test? The postgresql versions differ so I plan to just import the evergreen db instead of a dumpall. Will still have to edit a few things here and there but don't expect it to be painless. Second problem is BOBTFISH::Class-DBI-Frozen fails because it won't build without a working postgres server on localhost. I'm testing a split database so hadn't installed postgres and the dependencies coded in Makefile.install didn't catch it. Oops. Will install a server long enough to build it once I get past the more serious problem. Are you _sure_ it didn't install? I've never had to have a PostgreSQL server installed for that package to be installed. The CPAN directives in Makefile.install state that it should be installed forcefully - there are one or two tests that are known to fail and therefore kick out scary error messages, but they're red herrings. Apparently it did. After seeing the error I tried manually installing and couldn't and didn't think about the makefile doing a force. So that clears both holdups and I'm back in business. signature.asc Description: This is a digitally signed message part
Re: [OPEN-ILS-GENERAL] Problem with yaz, debian-squeeze and Net::Z3950-SimpleServer
Hi John: On Thu, Oct 6, 2011 at 7:24 PM, John Morris jmor...@beau.org wrote: Trying to install 2.0.10 and have hit an odd problem. Made it to install the prerequisites and can't do it. The problems seem to be two. The first is Net::Z3950::SimpleServer won't install from cpan. Squeeze doesn't have yaz and libyaz4 doesn't seem to You have to install a previous version of Net::Z3950::SimpleServer from CPAN. There were hopes that the maintainers would realize that it would be nice if their package would install on the most recent stable Debian release and make the dependency on the most recent source version of yaz optional, but that seems not to be in the cards. Net-SimpleServer-Z3950 1.12 works happily with Debian Squeeze. Don't have the command at my fingertips but something like the following from the CPAN shell should work: cpan install M/MI/MIRK/Net-SimpleServer-Z3950-1.12 be compatible. Although it does install a yaz-config as part of libyaz4-dev the cpan build process doesn't find it. Hacked around that and came up against the build failing to find yaz/facet.h and that file isn't on the system and dpkg -L libyaz-dev doesn't show it as part of it. So how has anyone else managed to install on debian? I'm assuming they have so what have I missed? Should I go find a tarball of an older yaz and manually build it? Really hate going outside the package system if I can help it. This is a relatively recent development (well - middle of August) and as the package is only required for providing a Z39.50 server it hasn't been on the front of people's minds; they generally don't run into it when testing out the basic functions of an installed system. We should be able to tweak Makefile.install accordingly - either to install the latest yaz from source (sigh - hopefully _that_ won't introduce problems to more core functions like z39.50 import...) or to install Net::SimpleServer::Z3950 1.12 specifically (I'd lean towards that because it has less impact on the system as a whole). Second problem is BOBTFISH::Class-DBI-Frozen fails because it won't build without a working postgres server on localhost. I'm testing a split database so hadn't installed postgres and the dependencies coded in Makefile.install didn't catch it. Oops. Will install a server long enough to build it once I get past the more serious problem. Are you _sure_ it didn't install? I've never had to have a PostgreSQL server installed for that package to be installed. The CPAN directives in Makefile.install state that it should be installed forcefully - there are one or two tests that are known to fail and therefore kick out scary error messages, but they're red herrings. -- Dan Scott Laurentian University
