On Fri, 2009-10-09 at 10:43 -0400, Mark Phippard wrote:
> On Fri, Oct 9, 2009 at 10:37 AM, Julian Foad <julian.foad_at_wandisco.com> wrote:
> > I found some strange behaviour in simple_gnome_keyring_first_creds() and
> > the three similar functions. If we try to unlock the keyring and the
> > attempt fails, the code says, "give up and try the next provider" and
> > returns no error but does not set (or save) the credentials. Is that
> > right? I think it should return an error in that situation.
>
> Suppose you have a client built with GNOME keyring support but you
> have never created a keyring and do not use it. Shouldn't SVN just
> continue to use the pre-existing password mechanisms, albeit with the
> new warning about plaintext? Why should it throw an error?
The desired behaviour is that Subversion should try the next provider,
yes, but that behaviour is obtained by the ..._first_creds() function
returning an error or setting its return value (*CREDENTIALS) to NULL.
This function does neither.
- Julian
------------------------------------------------------
http://subversion.tigris.org/ds/viewMessage.do?dsForumId=462&dsMessageId=2405573
Received on 2009-10-09 17:13:10 CEST