From: H. Peter Anvin Date: Tue, 11 Oct 2005 20:50:24 GMT Subject: Re: [RFC] Cleaning up die() error messages Message-ID: <434C2590.3040107@zytor.com> In-Reply-To: Matthias Urlichs wrote: > > IMHO, common library code should not be allowed to die. > (Yes, that does imply replacing all the xmalloc() calls.) > The sane way to do this is probably to call an overridable git_die() function, which can be specified by the user to use longjmp(), to use exceptions, or do something else appropriately. However, a much bigger problem is cleanup. -hpa