incubator-ooo-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Dave Fisher <>
Subject Re: [ISSUE] must have an MTA / MX
Date Wed, 02 Nov 2011 18:31:20 GMT
Joe is very busy, he is the individual who will implement the OOo MTA / MX.

I have learned that when he makes himself available it is time to ask questions. Asking questions
when he is busy requires patience and not too much repetition.

No matter which way the project chooses to go there are a couple of facts.

(1) MUST have an MTA and MX.

(2) Determining the Infrastructure requirements one way or another is necessity before we
can proceed.

(Choice one) If only a few addresses continue (like securityteam@oo.o) with the rest bouncing
with whatever message and link then Apache Infra can support it on the normal qmail/ezmlm
system. With or without preserving MLs as subscriber-less forwarders.

(Choice two) If many addresses are forwarded for any length of time then some variation of
choice two where the AOOo(i) PPMC must provide ongoing support for a non-highly available
mailing system on an Apache jail. This is true whether that is a dynamic system or a static

Rob clearly likes choice one (w/very few forwarders) and Dennis likes choice two (in a very
elaborate version.)

Sorry, if mentioning MLs caused anyone's head to explode, but please try to find what you
can agree on rather than focus on the one element you dislike and drown out the important
positive message.


On Nov 2, 2011, at 10:04 AM, Rob Weir wrote:

> On Wed, Nov 2, 2011 at 12:49 PM, Dennis E. Hamilton
> <> wrote:
>> Yes, I'll take my badge off and put my pistol in the safe:
>> Shane,
>>  1 (below). Your response fits the consensus I am seeing develop.  Nothing fancy,
bounce them, but some kind of catcher is needed to send something intelligible back other
than no-such-address.
>>  2 (below). I think killing all myname@ apart with limited forwardings
for committers it would be a horrible blunder of Klingon caliber: "All but committers are
unworthy worms."  It should make great slashdot reading.  Not to mention the glee on the faces
of those in the Apache AOOo Waiting To Fail (WTF) fan club.
>> Please do not collapse this system with the use of Apache IDs and myID@
addresses, even though there is a similarity in function.  This is more like what happens
when people register themselves to use the wiki or a forum or a mailing list, etc.  The complication
is that when folks registered on the OO.o site, they also got a forwardable e-mail address
that went with that ID.  And these ID/e-mail combinations are ubiquitous in the http:// *
ecosystem.  It is a kind of a fledgling, limited-reach OpenID/single-sign-on system.
> I think the chances are near zero that Oracle will give us the
> forwarding email addresses for 500,000 legacy
> addresses.   I also think it is clear that we are wasting time
> discussing this, time that could be spent making an unambiguous
> notification to those that have these addresses, allowing them to make
> an orderly transition to another address.  I further think it would be
> an unmitigated disaster if the forwarding service got shut down with
> little or no advance notice, because we wasted time discussing a plan
> that we will never be able to execute on.
> You might have different priorities than I have, but I hope you see
> the logic in the above.  If we're not getting the authoritative
> forwarding tables for the email forwarding, then we're wasting time
> right now and we should immediately start working on a notification
> for users currently using that service.
> If that is true -- and I think it is -- then the single most important
> thing we should be doing right now is establishing whether or not
> Oracle would provide such forwarding tables.  I thought we asked
> before and the response was "No".  But obviously that did not
> penetrate to an equal degree in everyone's mind.
> Would you agree that establish that simple fact is the most important
> thing to do now?
> -Rob
>>  - Dennis
>> I also have an addition to my list of where there are breakages with these e-mail
(not list) addresses: Some number of iCLAs have e-mail addresses as the e-mail
of record.  That won't interfere too badly with those who have already been established as
committers (at AOOo at least) but it is another bullet item for my original list.
>> -----Original Message-----
>> From: Shane Curcuru []
>> Sent: Wednesday, November 02, 2011 05:05
>> To:
>> Subject: Re: [ISSUE] Shut-down of all name@ e-mail addresses
>> I'll butt in with my (non-binding) suggestions.
>> On 11/2/2011 12:01 AM, Dennis E. Hamilton wrote:
>>> Hi Dave,
>>> where I am confused is the focus on Mailing-List forwarding rather than E-mail
>>> forwarding.  I can't tell what the intended behaviors are.
>>> Let's get clear:
>>>   1. If someone posts to one of the old e-mail list addresses (e.g.,
>>>, what is intended to happen?  What is the observed
>>> behavior?  How does this extend to use
>>> of -subscribe, -unsubscribe, -help, -owner (or their OO.o counterparts), etc.
>>> ?
>> After the final migration, they all bounce.  By then we should have nice
>> friendly pages - easily searchable for - that tell past
>> product version users which relevant Apache list to use.  And we'll have
>> sent several "hey, this list is going away" notes to the old list.
>>>   2. (a) If someone sends an e-mail to an existing account/e-mail address
>>> (e.g.,, what is intended to happen?  What does the
>>> individual that it current forwards to get to know or do about it?  The person
>>> sending the e-mail?  If the forwarding bounces, what will happen?
>>>      (b) If the account is closed/deleted, what are the 2(a) answers.
>> After the final migration, if they're not a committer, they bounce.
>> Period.  We are not in the business of providing services to non-committers.
>> If they're a committer, then it's up to the PPMC to decide if 1) you
>> want to, and 2) you will support some software to make committer
>> forwarders work somehow.
>> Apache projects use services to do their work.  I certainly
>> expect - as such a hugely accessed service - that web access to
>> will remain, along with a number of it's key subdomains.
>>  But we really need to start thinking like the new Apache project that
>> we're running, and not like some strange continuation of the past Oracle
>> project that is now... unsupported.
>> - Shane
>>>   - Dennis
>>> -----Original Message-----
>>> From: Dave Fisher []
>>> Sent: Tuesday, November 01, 2011 20:35
>>> To:
>>> Cc:; 'Joe Schaefer'
>>> Subject: Re: [ISSUE] Shut-down of all name@ e-mail addresses
>>> On Nov 1, 2011, at 8:07 PM, Dennis E. Hamilton wrote:
>>>> Whoa, now I am really confused.  This seems to have gone in the opposite
>>>> direction than what I thought.
>>>> First it narrowed down to privileging some small set of BZ users.
>>>> And then protecting our committers that have email
>>>> addresses.
>>>> Also, I don't think there had been any intention to preserve the
>>>> mailing lists.  Also, setting their addresses to forward to
>>>> different list that is not subscribed to is just weird.  So I don't
>>>> understand the list forwarding scenario.
>>>> And I have seen no one talk about moving the subscriber lists and adding
>>>> those subscribers to a list they did not opt into.
>>> Joe and I discussed doing it w/o subscriber lists. As a pure forwarder that's
>>> choice one and two.
>>>> I hope I misunderstand the common understanding about that.
>>> You do. See my other reply.
>>> [ ... ]

View raw message