[RFC] fileid involved tests want spaces in ids

[John Arbash Meinel]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Robert Collins wrote:
> I think these tests were just being aggressive about possible keys. I'm
> not aware of any key generation algorithm that actually created ids with
> spaces though. So I'd like to shrink their enthusiasm a little as I
> don't think we actually want to support file ids with spaces.
> 
> John, blame blames you - any comment?
> 
> -Rob
> 

We had bugs with different characters in file ids (most notably <>). So I wrote
a test which exercised as much as I could.

I'm +1 about file-ids not having whitespace in them, or even making them pure
ascii.

All bzr generated file ids will be pure ascii. (And won't have capital letters
in them since 0.9 or so).

We run into a bit of not knowing what will be present 'in the wild'.

So I'm okay with removing support for whitespace in file-ids, but I would
really like to see some strict tests about what we allow/don't allow.

To this point it has been more of a "what can we get away with", rather than a
strict definition.

So let's define it, write tests for it, and move clearly forward.

At this point, I would probably say that file-ids should be UTF-8 strings
without whitespace, and we encourage you to only use ASCII characters.

John
=:->

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.7 (Darwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFGrfz7JdeBCYSNAAMRAlWZAKDIPQcntGpljrMKL+dDRd8ZQ+1+SACfYM4j
uuyOtwGGigopCCoCG5rt50Y=
=r3Be
-----END PGP SIGNATURE-----