CAcert Code Development list.

[Wytze van der Raay <wytze AT cacert.org>]

Hi Eva,

On 02/21/2015 06:10 PM, Eva Stöwe wrote:
>> As far as I can see, both mails were sent to their intended audience.
>> The fixup of the queries was mainly "cosmetic", it did not affect the
>> real results. So unless someone can show this to be incorrect, there
>> is no need to send out another batch of emails.
> 
> the correct query would have been:
> 
> SELECT
>         u.id,
>         u.fname,
>         u.lname,
>         u.email,
>         COUNT(*) AS agreement_count
> FROM user_agreements AS ua
> LEFT JOIN users AS u ON u.id = ua.memid
> WHERE        ua.date >= '2015-01-08 14:29:00'
> AND        ua.date <= '2015-01-15 10:48:00'
> AND        ua.document = 'CCA'
> AND         u.id NOT IN (
>                 SELECT ua2.memid
>                 FROM  user_agreements AS ua2
>                 WHERE  ua2.date < '2015-01-08 14:29:00'
>                 AND   ua2.document = 'CCA')
> GROUP BY u.id;
> 
> The numbers who got the mail are quite hight, so I fear that we actually
> send a mail to the wrong audience.

Not so ... the above query will produce exactly the same results
as the one that we did use:

SELECT
        users.id,
        users.fname,
        users.lname,
        users.email,
        COUNT(*) AS agreement_count
FROM user_agreements
LEFT JOIN users ON users.id = user_agreements.memid
WHERE        user_agreements.date >= '2015-01-08 14:29:00'
AND        user_agreements.date <= '2015-01-15 10:48:00'
AND        user_agreements.document = 'CCA'
AND         users.id NOT IN (
                SELECT user_agreements.memid
                FROM  user_agreements
                WHERE        user_agreements.date < '2015-01-08 14:29:00'
                AND        user_agreements.document = 'CCA')
GROUP BY users.id;

> However we can send another mail to the same audience that got the first
> mail, to inform them about the mail-issue and that if they get another
> mail, that this would be correct, in any case.

Fortunately, that is not necessary (it would be quite confusing I think).

> I do not have an issue if that mail would be send in my name again. If
> needed I can just compose it - when you and Benny and maybe someone else
> review the text, this should be fine.
> 
> Afterwards we could send the first mail again with the correct above query.
> 
> (But please check the differences between the queries.)

There is a difference in formulating the query, but operationally they are
equivalent. Let's close the case ...

Regards,
-- wytze

Attachment: smime.p7s
Description: S/MIME Cryptographic Signature