Re: ntop Crashes Consistently on Mac OSX 10.5.6 after CryptoKey update
In article 250320091742442645%darrell.usen...@telus.net, Darrell Greenwood darrell.usen...@telus.net wrote: I find that ntop crashes consistently after SetCryptoKey occurs. iBook G4, 10.5.6, MacPorts 1.7, ntop 3.3.8. Log excerpts below. Any troubleshooting hints? Cheers, Darrell Mar 24 13:46:25 aabs kernel[0]: SetCryptoKey R: len 16, idx 2 Mar 24 13:52:06 aabs ReportCrash[48690]: Formulating crash report for process ntop[47893] snip Problem turned out to be flaky software in Airport Extreme Base Station. I changed Wireless Security from 'WPA/WPA2 Personal' to 'WPA2 Personal' and the problem was solved. Ntop has now run for 14 days without a crash. Cheers, Darrell -- To reply, substitute .net for .invalid in address, i.e., darrell.usenet6 (at) telus.net ___ macports-users mailing list macports-users@lists.macosforge.org http://lists.macosforge.org/mailman/listinfo.cgi/macports-users
Re: sysadmin: is there a manifesto on environment settings for macports?
Hi Rainer, I've had some difficulty with the VTK library when I try to get dynamic libraries working under macports. See http://trac.macports.org/ticket/19000 I could use some help with understanding how to get rpath working under macports (if that is THE way to go), given that VTK uses CMAKE settings to configure the RPATH or the INSTALL_RPATH and the settings that I've tried so far get corrupted during the macports build process because it uses a destroot setting. I've thought about a postdestroot hack using 'install_name_tool' but there may be a better way to get it right. For me, the bigger picture is creating an open-source scientific application that will depend on VTK, among other libraries, and macports may provide the foundation for the distribution process on OSX. Best, Darren On Thu, Apr 9, 2009 at 5:36 PM, Rainer Müller rai...@macports.org wrote: Darren Weber wrote: What about dynamic library path configuration (for dyld)? What is the best way to configure this for macports? Should settings apply at the system level (I would assume so) or at the user level (probably not). I currently have one env setting for postgresql83 (and I think this one is user specfic): DYLD_FALLBACK_LIBRARY_PATH=:/opt/local/lib/postgresql83 ports should not require any DYLD_* settings as they are expected to link against absolute paths in the ${prefix}. If this is really required for postgresql83 to work, it is a bug in that port. When building libraries, is there a general philosophy on using rpath? I'm confused about this, eg consider: http://wiki.debian.org/RpathIssue http://www.mail-archive.com/bug-libt...@gnu.org/msg00700.html http://lists.apple.com/archives/unix-porting/2008/Mar/msg8.html http://developer.apple.com/releasenotes/DeveloperTools/RN-dyld/index.html I've found that rpath can be a problem because macports builds into a destroot. When I use rpath outside of a destroot, it works fine, but I don't understand how to use it with a destroot, as in macports. If a port should use rpath, how should a port set RPATH correctly (maybe a relative path to ${prefix})? Or, should a port update a DYLD path setting so that run-time dependencies get resolved (see options for path settings in `man dyld` and maybe ports should set a system or user equivalent to ~/.MacOSX/environment.plist)? First, what particular problem are you trying to solve with this? Is there any port which relies on RPATH? If so, why would it need this? Rainer ___ macports-users mailing list macports-users@lists.macosforge.org http://lists.macosforge.org/mailman/listinfo.cgi/macports-users
Re: Readline fails to build (yet again)
On Apr 7, 2009, at 1:53 AM, Ryan Schmidt wrote: Attaching as a file is great, thanks; it ensures the output doesn't get reformatted by the email program. So, looking into this, the readline port tries to delete the file $ {prefix}/share/info/dir in post-destroot. It has done this since r7643 (2004-07-04). MacPorts base also deletes this file automatically, in all ports. MacPorts has done this since r7669 (2004-07-08). The readline port could be changed to delete the file only if it exists. Or, given that MacPorts base already does this, the line could be removed from the readline port. But I'm not clear why this is now, all of a sudden, almost 5 years later, causing a problem. BTW, I don't see the problem, on Mac OS X 10.4.11 Intel with Xcode 2.5 and MacPorts 1.7.1. Well, it was trying to install readline 6.0 (which is fine). However, I kept getting DESTROOT errors -- which I still don't fully understand (I'm in the midst of reading the DEV info now). So, I ended up removing all references to DESTROOT, namely the following: post-destroot { delete ${destroot}${prefix}/share/info/dir set docdir ${prefix}/share/doc/${name}-${version} xinstall -d ${destroot}${docdir}/html xinstall -m 0644 -W ${worksrcpath} CHANGELOG CHANGES COPYING NEWS README \ ${destroot}${docdir} eval xinstall -m 0644 [glob ${worksrcpath}/doc/*.html] \ ${destroot}${docdir}/html } # Install symlinks to avoid breaking ports linked against the old versions platform darwin { post-destroot { foreach f {history readline} { foreach v {0 1 2} { ln -sf lib${f}.${milestone}.dylib ${destroot}$ {prefix}/lib/lib${f}.5.${v}.dylib } } } } I'm not sure what problems that will present in the future, but it got past the errors. I'm still unclear as to where items such as DESTROOT and WORKSRCPATH are actually getting set. Thoughts on any of this? Thanks, Michael___ macports-users mailing list macports-users@lists.macosforge.org http://lists.macosforge.org/mailman/listinfo.cgi/macports-users