Why a Successful Migration Proves Nothing
A program can report on what it did. It cannot report on what you wanted, because it was never told.
Walkthroughs for the mail stores, archives and documents that ordinary software refuses. Written by the people who build the tools.
Showing all 39 articles.
A program can report on what it did. It cannot report on what you wanted, because it was never told.
Removing something properly means rewriting the file. Marking it as gone costs nothing. Formats built decades apart made that trade for the same reason.
Every individual step was sensible. None came with a note saying what it was, because writing that note was nobody job at the time.
A scanned page saved as PDF/A is a conforming file containing a photograph. It will look identical in fifty years and nobody will be able to search it then either.
A PNG showing block artefacts was a JPEG at some point. That is a reliable reading whatever the extension now says.
The page is designed to look like paper and the proof is not on the page. That is the whole of it.
A black rectangle over text is a reasonable idea of what redaction means, and the format quietly does something else.
The page renders from the font and copying reads a separate mapping. Nothing forces the two to agree.
Storage is cheap and storage is not the cost. Everything kept is everything that can be asked for.
Almost every surprise here comes from somebody believing they did the fifth step when they did the second.
A run that moved everything and broke every attachment reports a perfect total. The count is the one thing that proves nothing.
Marketing email is built with tables and inline styles because that is the only construction that survives every client.
Half the mailbox an hour out and half correct reads like corruption. The split falls exactly on the date the clocks changed.
The people holding DBX files in 2003 had no reason to worry. It was the most common mail format on earth.
Maildir is better while mail is arriving. MBOX is better once it has stopped. That split is not accidental.
A quoted value may contain a line break. It is also the one thing most simple parsers get wrong, and mail exports produce it as a matter of course.
The error message never mentions line endings. It names a file that plainly exists and appears to be spelled correctly.
No key means no secret. If nothing had to be known to produce the result, nothing has to be known to undo it.
A wrong extension is a labelling mistake, not damage. The bytes were never touched and renaming costs nothing.
The bytes are almost always fine. What broke is the agreement about how to read them.
A privacy policy is a promise. A tool that finishes the job with the network switched off is a demonstration.
Most people never chose. A setup wizard chose for them. The consequences turn up when a laptop dies or an account closes.
The format that keeps more is also the one that fewer things can read. That trade is the whole decision.
The organisation believed it had seven years of mail. It had fourteen days, repeatedly, for seven years.
A handful of new headers, all of them plain text, all of them safe for a server that had never heard of them. It asked permission from nobody.
The same structure on disk, holding the same things, under two different arrangements about who it belongs to.
Received lines are written bottom upwards, so the oldest hop sits last. Read from the bottom and you are following the message forwards.
A stated limit counts the message rather than the file. Twenty five megabytes of message holds about eighteen megabytes of anything you attach.
Outlook picks a message format per recipient rather than per message, which is why the same sender reaches ninety nine people cleanly and one badly.
A full DBX refuses new messages and refuses deletions too, so the first thing everybody tries is the one thing the file will not allow.
Every value sits in its own stream named __substg1.0_0037001F or similar. Nothing in the file says which of those is the subject line.
A saved page carries a Subject line and a sender because the container expects them. Neither means the file was ever sent anywhere.
Dirty shutdown reads like an error and is not one. Every database currently serving mail is in that state, which is why the header tells you so little.
The file copies perfectly. Outlook on the other machine is what refuses it, over one record written inside the file.
Rename it to .zip and it opens. Inside, every message has been rewritten as XML, which is why getting mail back out is a rebuild.
The format has no version and no variants. Everything awkward about EML comes from the folder around it rather than from anything inside the file.
Plain text, no index and no specification for its first thirty years. That history explains the four variants, the size on disk and why Outlook will not touch it.
Rules, views, categories and search folders all live in there beside the mail. None of them survive a move to any other format.
Two questions sort all of them. One message or a whole mailbox. Published structure or private. Everything else about portability follows from those two answers.
Every tool runs inside the browser tab, so a file is read where it already sits and nothing is sent to a server.