Re: Do I need multiple publicinbox.<name>.address values?

[ebiederm@xmission.com (Eric W. Biederman)]

Eric Wong <e@80x24.org> writes:

> "Eric W. Biederman" <ebiederm@xmission.com> wrote:
>> 	my $tracker = PublicInbox::IMAPTracker->new();
>
> Thanks.  What's PublicInbox::IMAPTracker?

Something that keeps the last fetched UID in an sqlite database.
I will follow up with a patch for that as well.

I haven't been brave enough to let this script delete any mail
yet so I need to track what has been fetched.  Something that
will be rolled back if the email message isn't commited into git.

I have a companion script that will delete mail.

I mostly sent it so that there is some idea what I am working with.

In context of this discussion I don't remember how often I am looking at
the "Delivered-To:" header.  That is almost gone but I know for a while
I was using that as well.

>> for (;;imap_sleep()) {
>> 	# Run fetch_mail in it's own separate process so
>> 	# that if something goes wrong the process exits
>> 	# and everything cleans up properly.
>> 	#
>> 	# Running fetch_mail in an eval block is not enough
>> 	# to prevent leaks of locks and other resources.
>
> I was thinking along the same lines :)
>
> The rest looked good and it's definitely something I can work
> with to get supportable in a way that reuses git's credential
> system and configs.

I think fundamentally the script sucks because it is one email message
at a time (not using IMAPs overlapping), the script is polling (NOT
using IMAP idle), and I have magic to go from the server that I need
to fetch to the actual domain.

But in practice the script seems to be working reliably even in the
event of an error right now.

Eric