http://issues.apache.org/SpamAssassin/show_bug.cgi?id=4635
[EMAIL PROTECTED] changed:
What |Removed |Added
----------------------------------------------------------------------------
Attachment #3195 is|0 |1
obsolete| |
------- Additional Comments From [EMAIL PROTECTED] 2005-10-25 06:44 -------
Created an attachment (id=3205)
--> (http://issues.apache.org/SpamAssassin/attachment.cgi?id=3205&action=view)
Updated patch to avoid double-connect
(In reply to comment #3)
>
> I believe you are mistaken in how the code works. That said, it's been
> awhile so it's not coming to me immediately. I'll have a look and see
> just to make sure.
>
I've verified that if tie_db_readonly() is called, followed by
tie_db_writable(), and the user has no bayes information yet, two seperate
database connections are performed: once for _initialize_db(0), which
disconnects as soon as _initialize_db(0) "fails", then again when
tie_db_writeable() is called - this is easy enough to test by setting
$ENV{DBI_TRACE} = 1 before SA is loaded, and looking for ->connect's in the
resulting trace debugging.
My previous patch was broken, however, in that it needed to call
_initialize_db(1) if $self->{db_writable_p} is false just before it sets it to
true in tie_db_writable().
------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.
