[BUG][PATCH] fileid_involved fails to unescape xml characters

[Martin Pool]

On 19 Jan 2006, John Arbash Meinel <john at arbash-meinel.com> wrote:

> Do you have a specific timeline for releasing 0.7rc3? The
> fileid_involved bug basically breaks everyone who uses an arch->bzr
> conversion, since directories generally got a tag which was:
> 	User <email> Timestamp
> Which means there are lots of <> characters in Arch ids.

I'll make one it in the next 2 hours.  

For people who've suffered this I think it will be enough to just remove
the files from inside revision-store describing newly pulled in
revisions whose content didn't come across.  The problem here is this
code introduced in Goffredo's patch:

            revs = self._revids_to_fetch(last_revision )
            # nothing to do
            if revs:
                self._fetch_revision_texts( revs )
                self._fetch_weave_texts( revs )
                self._fetch_inventory_weave( revs )
                self.count_copied += len(revs)

It breaks an invariant to pull the revision texts into the store before
their content has come in: if the transfer is interrupted then the
revisions will be half-present, which is not allowed.

It would perhaps be nice to be able to fetch them into an in-memory
isolated transaction and write them out afterwards but that's not what
happens at the moment.

(We should perhaps add a test with a transport that interrupts after a
certain number of requests to exercise handling of flaky connections.)

-- 
Martin
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
Url : https://lists.ubuntu.com/archives/bazaar/attachments/20060120/5480c711/attachment.pgp