Re: gc.m4 and hard failure

[Paul Eggert]

Simon Josefsson <address@hidden> writes:

> Now that this is settled, I'm beginning to think that the random
> functions should be decoupled from GC fully.  I.e., there should be a
> standalone module 'random', and then gc-random could use it.  This is
> more similar to other gc modules.  IIRC, CoreUtils random-sort might
> be one user of this 'random' module.

Something like that sounds reasonable.  I have some random code
already drafted for coreutils, designed to be used for both 'sort' and
for a new 'shuf' program.  I'll take a look at gc* and see how well
the two modules compare.