[RFC] fileid involved tests want spaces in ids

[Robert Collins]

On Mon, 2007-07-30 at 09:36 -0500, Martin Pool wrote:
> 
> I don't think we generate spaces with ids either.
> 
> I just saw more posts on the bug about supporting '\n' in filenames.
> It
> is a little different in that we have no choice about filenames (when
> importing), but we can reasonably refuse some ids.
> 
> However, I do think whenever possible we should accept all data of the
> relevant type, even if we restrict it by convention.  If we later
> discover
> that we do need to support them then it is more trouble to change the
> format later on.  I'd want to see some really compelling performance
> or
> simplicity reason to restrict it.

fileids make up partof the composite key used to access file data. Right
now a file id with a space in it generates a knit file on disk with a %
20 embedded. I think its easier to be able to assume that whitespace is
not part of the keyspace of internal data - and as this is ids, not
paths, it doesn't have to be - than to allow any whitespace in this
data. John has confirmed that there was not a specific case of ' ' being
used that he is aware of, and we can search the supermirror for ids
matching this too to be sure.

If we do find this is too restrictive in the future, I'd be happy to be
the one to do the work to loosen up the index to support whitespace in
keys.

> In other words I don't think you should remove that test. 

I'm not planning on removing it, just taking the ' ' out of it.

-Rob

-- 
GPG key available at: <http://www.robertcollins.net/keys.txt>.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part
Url : https://lists.ubuntu.com/archives/bazaar/attachments/20070731/31769783/attachment.pgp