On Tue, Jul 15, 2008 at 11:39:12AM -0400, Karl Fogel wrote:
> stsp_at_tigris.org writes:
> > Log:
> > * subversion/libsvn_subr/simple_providers.c:
> > (svn_auth__simple_save_creds_helper): Test a pointer pointing
> > to a boolean against NULL explicitly. The code as written made
> > it relatively easy to confuse the pointer with the boolean itself.
> > No functional change.
>
> +1 on this change, but just a heads up, stsp: I'm planning to rewrite
> that code to use a plain boolean (instead of a pointer to boolean), for
> simplicity. You can see how it will look by examining:
>
> http://subversion.tigris.org/nonav/issues/showattachment.cgi/907/2489-patch-v5.txt
>
> which has similar code (e.g., same variable names) in another file.
>
> (I don't want to change simple_providers.c until I've committed the
> above patch, which is undergoing 'make check' right now.)
You mean this bit, right?
+ /* We're interactive, and the client provided a
+ * prompt callback. So we can ask the user.
+ *
+ * Check for a cached answer before prompting. */
+ svn_boolean_t cached_answer =
+ (svn_boolean_t) apr_hash_get(b->plaintext_answers,
+ realmstring,
+ APR_HASH_KEY_STRING);
+ if (cached_answer)
+ may_save_passphrase = cached_answer;
+ else
+ {
+ /* Nothing cached for this realm, prompt the user. */
+ SVN_ERR((*b->plaintext_passphrase_prompt_func)
+ (&may_save_passphrase,
+ realmstring,
+ b->prompt_baton,
+ pool));
Looks good!
Stefan
- application/pgp-signature attachment: stored
Received on 2008-07-15 18:52:06 CEST