On 10/31/2011 2:08 PM, Sandro Tosi wrote: > I'm considering reverting the change you mentioned, but you have to be > more convincing than before.
Sure, that's fair. For clarity, I would prefer adding an additional configuration parameter rather than requiring the envelope sender to be either the from address or system default, because I think the submitter of bug #614880 has a reasonable use case. But making that change requires supporting arguments as well, so they are provided below. Note that in the cases below the alternative to setting an envelope sender configuration option would be configuring reportbug to use SMTP. I contend that this raises the barrier to entry over using a local MTA when one exists because the user must deal with ISP or local outgoing port blocking, figuring out appropriate SMTP options (e.g. GMail's non-standard configuration requirements), and configuring them for yet another program on all systems used. If you disagree, the scenarios below will likely not be persuasive. The arguments also depend on the extent to which ease/convenience is a goal of reportbug. Feel free to consider them with that in mind. > On Thu, Sep 22, 2011 at 20:58, Kevin Locke <ke...@kevinlocke.name> wrote: >> I contend that it is not reasonable to expect users to have local MTAs >> configured to canonicalize/masquerade local user account names to public >> addresses > > could you please elaborate why it's not reasonable? Sure, here are some use cases that I contend reportbug should support: * Reporting bugs from systems on which the reporting user does not have administrative access to configure the MTA or the address mapping for their user account. * Users who either lack the knowledge to configure their MTA, prefer not to make changes to a working system, or prefer not to spend the time required to determine and implement the correct configuration to satisfy reportbug. >> and that there are legitimate cases where users will want to >> report bugs with addresses different than whatever their default >> envelope sender is. > > examples? Here are a few example cases that I would consider reasonable: * Suppose the user has separate professional and personal email accounts. Their MTA is set to send as one of these accounts but they may want to report the bug using either their personal account (if the bug is not work-related) or professional account (if it affects systems they maintain professionally). * To ease the sorting of messages, the user would prefer to report bugs using <name+...@domain.tld> to automatically tag responses "bts" but would not want to set their address for all mail as <name+...@domain.tld> or to change the address each time reportbug is used. * The user is logged into a system as a contractor and the configured email for their system is either temporary (e.g. for the duration of the contract) or an email they prefer not to use for the purpose of submitting bugs. >> Also that the From header should match the envelope >> sender, where possible. > > references to back up this assertion? This was poorly phrased on my part. It's not that all mail should have the From header == envelope sender, but that when the envelope sender is not set to the From address it opens the door for problems on some systems. If the MTA is not configured with a valid email address for the user the envelope sender will be sent as a locally-qualified address (e.g. user@hostname) which will be rejected by the BTS.[1] Hopefully that explains my thoughts somewhat. If you would like a patch, let me know. Cheers, Kevin 1. If a masquerade domain is set (so mail will be sent as user@masquerade) in which the username does not receive mail it may be rejected by sender verification depending on how the BTS is configured. I can test this if you like. _______________________________________________ Reportbug-maint mailing list Reportbug-maint@lists.alioth.debian.org http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/reportbug-maint