[jira] [Commented] (SVN-1804) mailer.py doesn't catch SMTPRecipientsRefused in finish()

["Nathan Hartman (Jira)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/SVN-1804?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16963607#comment-16963607 ] 

Nathan Hartman commented on SVN-1804:
-------------------------------------

This issue is fixed in r1869194.

To fix this issue, we now handle SMTP exceptions.

When an exception only affects a particular email, such as invalid recipient, the error is reported to stderr and execution continues to avoid losing any later emails.

In contrast, fatal SMTP errors, such as login with bad credentials, terminate execution as before, but with some additional reporting to stderr.

As before, the script's exit code is zero if all emails were sent successfully, nonzero if any SMTP error occurred.

> mailer.py doesn't catch SMTPRecipientsRefused in finish()
> ---------------------------------------------------------
>
>                 Key: SVN-1804
>                 URL: https://issues.apache.org/jira/browse/SVN-1804
>             Project: Subversion
>          Issue Type: Bug
>          Components: tools
>    Affects Versions: all
>         Environment: Windows 3.1
>            Reporter: Subversion Importer
>            Priority: Major
>             Fix For: nonblocking
>
>
> {noformat:nopanel=true}
> I haven't decided if this is a bug or a feature yet.  I can see some advantages to either method.  That 
> said, I'm adding it as an issue anyway because it makes generic project type handling problematic.
> finish() of mailer.py does not attempt to catch the exception SMTPRecipientsRefused, which is raised if 
> the recipient of the destined mail does not exist.  This blocks any other mails that might be sent out as 
> well.  Given the commit has already happened, and the lack of error/message reporting of the hook 
> system in general, my solution was just to catch the exception and pass.  I'm not sure whether or not 
> I'm satisfied with that solution though, as a message should probably be sent to stderr.
> {noformat}
> Original issue reported by *rzigweid*



--
This message was sent by Atlassian Jira
(v8.3.4#803005)