running irssi with any parameters from gentoo minimalist cd (2.6.17-gentoo-r7) fails, issuing an error: NO WINDOWS: -!- Irssi: Invalid charset: ISO8859-1 *** glibc detected *** free(): invalid pointer: 0x080ea397 *** Aborted (core dumped) Whenever I supply any paramters (such as -n <nick> or -c <connection>, regardless of order, I receive a messsage such as this and it crashes. Without parameters, it only prints the first line of the error prior to opening the irssi screen. However, in here I can use icq commands to connect and chat as I wish. I've been playing with various installs, deleting my entire hard drive and restarting often. This error appears prior to chroot'ing. Each time but once that I've booted up and tried to run this, I've received this type of error; however, the charset is occasionally 1259 instead of 8859. Reproducible: Sometimes Steps to Reproduce: 1. boot into into minimalist cd 2. at the console, type irrsi -n <nick> or -c <conn> or both
When I run it, I get the "Invalid charset" message, which sometimes changes which charset it's complaining about, but irssi runs just fine. Try the CD in another system, try to burn the CD again, and try running memtest86+ on your machine. Please reopen if you try all of these and it still happens.
> Steps to Reproduce: > 1. boot into into minimalist cd > 2. at the console, type irrsi -n <nick> or -c <conn> or both > The same for me - I get segfault, running irssi without parameters just works fine, but using e.g. /set nick woodyPL causes crash too. I remember not having this issue on x86 - only on amd64...
Can you guys verify this issue with 2008.0 Beta 1?
Just verified it with beta2. I'm trying to get a copy of the log atm.
Created attachment 153089 [details] Log from attempt to run irssi on 2008.0_beta2 x86 minimal CD
I'm CCing the irssi maintainer. Any ideas?
I can reproduce this and know what the culprit is. The "Invalid charset: ISO8859-1" comes from not having /usr/lib/gconv on the livecd, Irssi then falls back to a hardcoded charset name, which is a constant string, but later gets freed and then causes the abort.
This issue has just been fixed upstream and I pulled the patch into our package.
OK, this is now updated in the release snapshot.
This should be fixed in 2008.0