Re: [Monotone-devel] RFC: Fake IDs

[Florian Weimer]

* Zack Weinberg:

> Typical use case (from cmd_merging.cc - I'm not posting a diff for
> that, as I have a lot of other changes in that file right now):
>
>       // Just pick some unused revid, all that's important is that it not
>       // match the work revision or any ancestors of the base revision.
> -      r_target_id =
> revision_id(hexenc<id>("5432100000000000000000000500000000000000"));
> +      r_target_id = revision_id(fake_id());

Can't you use a tagged union of an SHA-1 hash and a fake (but
guaranteed to be unique within the ssion) ID?  On 64 bit
architectures, this shouldn't result in an observable space
difference.