That's still strange to me, in that Fink's "tail" comes with coreutils-default, and I had that installed when I did my 10.4 build.
Silencium Sixty-Eight wrote: > Hi guys, > > I've tracked it down: when using "tail" provided by Fink I can't compile net-snmp-unified! :( > > After removing it and using the stock MacOS X one, everything went just fine! :) > > Cheers, > Martin > > On 13.12.2007, at 19:28, Alexander K. Hansen wrote: > > Martin Costabel wrote: > >>> Alexander K. Hansen wrote: > >>>> -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 > >>>> > >>>> Silencium Sixty-Eight wrote: > >>>>> Hello Benjamin, > >>>>> > >>>>> I'm trying to setup the fink package "net-snmp-unified-5.4.1-1" > >>>>> on my MacOS X 10.5 box; unfortunately it fails to succeed with > >>>>> the following error message: > >>>>> > >>>>> making all in > >>>>> /usr/local/src/fink.build/net-snmp-unified-5.4.1-1/net- > >>>>> snmp-5.4.1/agent Makefile:218: *** commands commence before > >>>>> first target. Stop. > >>>>> > >>>>> When investigating the agent's Makefile I discovered that the > >>>>> trailing backslashed in the "mibgroup_list_lo", > >>>>> "mibgroup_list_o", "agentgroup_list_lo" and > >>>>> "agentgroup_list_o" definitions where missing. > >>>>> > >>>>> After adding them I could compile the agent, but another error > >>>>> occured with the mibgroup Makefile (just the same, missing > >>>>> trailing "\"). > >>>>> > >>>>> Cheers, Martin > >>>>> > >>>> I would be curious whether other people on Leopard see the same > >>>> issue. The package built properly for me on Tiger. > >>> > >>> On Leopard/ppc I don't see it. > >>> > >>> Those Makefiles are constructed by a complicated recursive > >>> procedure in configure, using low-level text processing tools like > >>> 'sed' and 'echo'. > >>> > >>> I would suspect first 'sed'. If a sed from Fink is installed, or > >>> another one is on the PATH except /usr/bin/sed, I would remove it > >>> and see if this changes anything. > >>> > >>> The second suspect would be 'echo'. I see that the configure script > >>> of net-snmp now has a tricky check whether 'echo' understands the > >>> -n or \c options. Good thing, in principle, seeing that Leopard has > >>> several versions of echo that behave differently. But this check > >>> does not seem to work correctly. In my case this didn't matter, but > >>> it could be different in other situations. > >>> > >>> I say it doesn't work correctly, because the result of the test is > >>> a nonsensical line > >>> > >>> ECHO_C='ECHO_N='' > >>> > >>> in config.log. A LF character got lost somewhere between configure > >>> and config.log. Maybe the finally chosen 'echo' or $(ECHO) does not > >>> understand configure's wish to output backslashes. > >>> > It looks to have built OK with fink's sed on 10.4, too. >> >> - ------------------------------------------------------------------------- SF.Net email is sponsored by: Check out the new SourceForge.net Marketplace. It's the best place to buy or sell services for just about anything Open Source. http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace _______________________________________________ Fink-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/fink-users
