On Thu, 19 Jun 2003 12:30:48 -0400 "Amy A." <[EMAIL PROTECTED]> wrote:
> > > I did a little checking and found that my evolution could be crashing > > > due to a galeon bug. My problem seems to be related to bug 4071, but > > > reading the bug report it does not say how to reslove this issue or list > > > any action taken. If anyone can point me in the right direction, I would > > > greatly apprecitate it!!! > > > Amy A. > > > > > > [EMAIL PROTECTED] bin]# urpmi evolution > > > To satisfy dependencies, the following packages are going to be > > > installed (23 MB): > > > evolution-1.4.0-1mdk.i586 > > > gal2.0-1.99.7-2mdk.i586 > > > gtkhtml3.0-3.0.5-2mdk.i586 > > > libgtkhtml-3.0_2-3.0.5-2mdk.i586 > > > Is this OK? (Y/n) y > > > installing > > > > [...] > > > > > [EMAIL PROTECTED] bin]# evolution > > > evolution: relocation error: evolution: undefined symbol: > > > ORBit_c_stub_invoke > > > [EMAIL PROTECTED] bin]# > > > > You should not install cooker packages on a 9.1 install. You should run > > full cooker, or run a stable release. > > Instead of running "urpmi evolution" run "urpmi --auto-select" which > > should upgrade everything to the latest in cooker (after you updated the > > hdlists). > > > I will certainly give that a try, but one of the advantages of urpmi it that > > it tells you the deps that will be needed. I have not heard that I should > only run a 'full' cooker box and not be able to choose a specific rpm only. > It's not clearly written in the cookerdevel page or in the cookerfaq, true. The message does come by on the mailinglist regularly though. It is written on those pages that you should search the archive first, before reporting, and some days ago it was mentioned that you need to upgrade bonobo and orbit, otherwise gnome will break. So you could have known :-) (evolution is a gnome package). > I found using urpmi to be much more effective for installing since it grabs > my deps also (except for this time ;-) Yes, but it can't grab every dependency. Sometimes things just break, like when a package is rebuilt with a new compiler (happened with gcc2 => gcc3 a lot). The maintainer of the package would need to set this manually, which will become unmaintainable very soon. Also, when using a mix of older and newer packages, it becomes very hard to determine what still works and what doesn't. There are millions of possible installs then. You can't test all these possibilities. When you follow the rule to run a stable release like 9.1, or to run full cooker, the number of possible installs becomes a lot smaller. I'm not sure if I wrote that in understandable words :-) -- Marcel Pol
