[MERGE] Make ftp:// a little more resilient for initial push

[John Arbash Meinel]

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

Martin Pool wrote:
> On 6/5/07, John Arbash Meinel <john at arbash-meinel.com> wrote:
>> The former biases "put_file()" to assume it is failing because the
>> parent dir
>> doesn't exist, and the latter adds a specific string that we have seen.
> 
> The second certainly makes sense.  The first one I'm not so sure,
> perhaps it will just cause further confusion if something else is
> happening?
> 
> I would think we only really expect put_file to fail because the
> parent doesn't exist in certain particular situations (like making a
> new branch?) and maybe it should be explicitly handled there.
> 
> But if you think it's an improvement I don't object.
> 

It is actually how we do it for sftp, because the common error is parent
does not exist. Mostly because the only times we "put" are when creating
a new file (because we know the knit didn't exist when we tried to open
it), or replacing a file (.bzr/branch/revision_history).

Otherwise we would be appending to an existing file.

Because we are doing it with sftp, it seems reasonable to do it for ftp.

John
=:->

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.3 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFGenyEJdeBCYSNAAMRAo5nAJ9FuEXh1d99ytEgJH02arLwEjtLRQCfaaJD
Uocp7JaDkNNujJyosUXYLEM=
=xiwR
-----END PGP SIGNATURE-----