Re: Problems with byte compiled startup files

Mike FABIAN writes: > http://tracker.xemacs.org/XEmacs/its/issue405 > > which was reported as > > http://bugzilla.novell.com/show_bug.cgi?id=432404 I can't replicate with your example init file (does the custom.el file actually exist?

I don't have one in my test set up since you didn't mention its contents) on any of several platforms (PPC and Intel Mac with Mac OS X 10.4 and 10.5 respectively, Gentoo Linux/AMD Opteron, NetBSD 3/Intel Pentium III), and I have always (since I started using XEmacs in 1996) byte-compiled my init file. I think most of the developers do. I've never heard of a problem with byte-compiled init files, whether init.el or application dotfiles. My tests are with Mercurial tip or close to it, though. I will try with a vanilla 21.5.28 tomorrow, I guess, but I don't think any changes have been made to the byte-compiler or byte-code interpreter recently (Aidan seems to have been fiddling with it in Dec 2007). One possibility is that you've got an evil C compiler. I can't help with that because XEmacs currently compiles warning-free with gcc, except for the warnings about "not a prototype" in getopt.h, the attempt to fool the compiler into accessing memory in a debug function, and a deprecation warning in a Mac OS X header file. That is with the default GCC 4.0.1 (Apple build) and GCC 4.3.2, and with g++ of GCC 4.3.2 (except that g++ complains about implicit conversions from const char[] to char* a few hundred times, but that seems unlikely to be related to this). Another possibility is that it's related to 64-bit uncleanliness, but that's something I know nothing about, and my Opteron box is in 64-bit mode.