Mark,

This might have an added benefit for your servers, reducing needed
storage. If it did, it would keep costs down for the platform.

Donald

On Sun, 24 Dec 2017 13:31:38 -0800, "Shal Farley" <shals2nd@gmail.com>
wrote:

Mark,

A couple group admins in GMF have noticed that their groups' Emailed
Photos albums have become stuffed with multiple copies of images from
members' email signatures - over 670 in one case, over 36,000 in another!
https://groups.io/g/GroupManagersForum/topic/7704553

The fact that the images are shown in the Emailed Photos album is a
problem in that it clutters that album to the point that it is useless
for finding other photos.

Another concern is the storage used. My belief is that having them
listed in the Emailed Photos album does not itself consume any extra of
the group's storage quota. That is, the storage was allocated to
preserve the image as seen in the message in the Messages archive;
showing it also in Emailed Photos album did not require an additional
copy of the image. That said, and even given that images in signatures
tend to be small files, if a group accumulates many thousands of them
the storage cost could add up beyond insignificance and might eventually
risk putting the group over its quota.

The first response was a wish that bulk delete had already been
implemented. cf:
https://trello.com/c/DtJTcUrg/233-better-way-of-bulk-deletion-of-photos-both-of-emailed-photos-and-regular-photos

However, in the face of 36,000 copies even that would be insufficient
unless it has an automatic way of selecting just the photos matching one
of them.

*1) Remediation. *Provide a means to manually select one photo in an
album and have the system find and select all matching photos as targets
of a subsequent a bulk action.  Note, for this purpose I believe an
exact match of the image file to be sufficient - it isn't necessary, or
even desirable, to have a perceptual (or "looks like") match.

*1a) Action: delete.* I'm imagining that this action must not only
remove the image files, but also their slots in the Emailed Photos album
and do something with each source message to avoid leaving a broken
reference.

*1b) Action: consolidate.* This would be more subtle. This would leave
behind a single copy of the image file, deleting the rest. I imagine
that the source messages would have to be updated to point to this
single copy in place of the deleted copies.

Handling this case in the Emailed Photos album is a bit trickier.

Where does the one remaining album entry sort? For Posted date the
obvious choices are: at the date of the manually selected target, at the
date of the oldest copy, at the date of the newest copy. I'm inclined to
take the oldest copy as the best of those choices. For sort by Taken
date they copies would perforce all have the same Taken date - or more
likely no such date at all (no EXIF data). Name sort is trickier, as the
names may be somewhat arbitrary, but again I think the name of the
oldest copy is likely a good enough answer.

Also, what of the Message button? Would it in any scenario be useful for
that to invoke an advanced message search for all the messages that
contain the consolidated image?
I can almost imagine a group admin wanting a way to find them all. If
not that, then I imagine having it bring you to the oldest message would
be the best answer.

*2) On-going maintenance.* Whatever remediation (1) is taken, the group
admins still face an on-going chore of repeating the consolidation or
deletion on a regular basis; as fresh messages cause the images to build
up again.

To avoid this, it would be desirable for the system to compare the
images in incoming messages against those which have been consolidated
(1b) and consolidate the new image as well. If the action taken was
deletion (1a) there isn't a remaining copy to compare against the
incoming images, and it isn't clear that the group managers would want
to set up an automatic deletion in any case.

*3) Fully automatic.* As an alternative to (1), it would be possible to
automatically detect a repeated image file as each new message is
received. This, combined with a one-time pass over the existing message
archives, would make the consolidation feature completely automatic,
requiring no effort from the group admins.

The matching could be restricted to messages from this member, to reduce
the search space and keep the added processing time to a minimum.

Shal

The further a society drifts from truth the more it will hate those who
speak it. --George Orwell