Hardlink Dupes

[Mathias-Ro]

Hello,
often the same Mail is in different folders (different mail accounts at the server).
Here it would be nice to replace the mails with hardlinks.
Best regards,
Mathias

[leggewie]

Since you are mentioning hardlinks, I will assume you refer to maildir settings only. For mbox this request is obviously impossible and unlikely to be what you had in mind. The problem then comes down to a) files are identical (same hash from a filesystem perspective) and b) files are identical for the purpose of mdedup but not for the FS.

I'd think that implementing the requested feature is likely feature creep with a lot of potential cornercases. In situation a) you are better off with existing lint binaries (I maintain fslint in Debian, that's 1 option, another is duperemove or hardlink). Just choose what suits you best. I'm not sure that b) is enough of a problem to justify to tackle it.

[Mathias-Ro]

If the same mail is stored in different boxes the mails have different x-headers, so they are no dupes for linux. For the user it's the same to see his own sent mail or a hardlink of the received mail in his sent folder.
If there are forwarding rules all could use hardlinks of the same mail file.
With a little shell script I found in different organisations a medium of about 2 till 4 identical mails in different folders.

[leggewie]

I'm not saying there is no use case for this. I'm just not sure there is enough of a use-case for mdedup to add this feature. If as you say "For the user it's the same to see his own sent mail or a hardlink of the received mail in his sent folder." then I have to wonder why the user is sending CC: to himself. I don't, for that exact reason to avoid unnecessary dupes. Others might want the dupes and want to keep them.

I'm just worried that this adds very little benefit and lots and lots of new cornercases and things to consider. Hardlinking files that are not identical brings back memories of the Xerox workcentre scandal. It's one thing to drop files. It's not the same to subtly and undetectably alter stuff. The more I think about this idea, the bigger my reservations.

[Mathias-Ro]

If as you say "For the user it's the same to see his own sent mail or a hardlink of the received mail in his sent folder." then I have to wonder why the user is sending CC: to himself.

In imap folders the mail client stores a copy of the sent mails in the .sent folder of the acccount. It's the only way to see the sent mails at all his devices.

Some users or organisations use mails rules for some copys of the same mail inside the organisation. Here we also could replace the copies with hard links.