Trying to upgrade the ancient glibc-2.10 that comes in Red Star OS with glibc-2.23 and instead of working, I broke the system spectacularly. 😂

Maybe I shouldn't have used /usr as the prefix for configure? Oh well, lessons learned! Time to reinstall!  

#RedStarOS #ThereWasAnAttempt #HackThePlanet

@lambdacalculus Hi, uh, I moonlight as a glibc developer sometimes. The trick for doing this successfully by hand is, configure with --prefix=/usr (if you don't it breaks the ABI) but then use "make install DESTDIR=..." to install it to a staging area, and then *reboot into a liveCD or something*, manually mount the partitions, and copy all the files into place by hand (there's a way to do it with one invocation of rsync but I don't remember the exact command offhand).

@lambdacalculus the reason a direct "make install" breaks everything is, to replace the active copy of glibc, libc.so.6 and ld-linux.so.2 and a couple other files have to be replaced *in synchrony*. Package managers can do this, but "make install" can't. If they get out of sync then you have the old dynamic loader trying to load the new libc.so, or vice versa, neither works.

So if you don't have a package manager to help, you gotta do the operation with the help of a separate C runtime.

@zwol I did use '--prefix=/usr' when running configure, but the other thing is that Red Star OS is, for lack of a better description, held together with paper clips and chewing gum. North Korea, in their quest for "security systems", broke so many aspects of Fedora (the base of Red Star 3.0) that breathing funny on it will cause the system to shit the bed.

Hilarious that I can modify most everything *except* glibc, but I'll see if I can try your advice in a VM as a tester.

@lambdacalculus Right, understood. I don't know anything about red star but I have seen exactly the failure you hit with glibc before, is all :)