For all I care (and I guess the down-voters in your case think similar) that's just the same thing, really.
If you can reproduce the plaintext in any way you're guilty of storing plaintext passwords. Arguing about the terms doesn't change the problem nor the perception in this humble author's opinion. The article is ~correct~, good enough, works for me.
> If you can reproduce the plaintext in any way you're guilty of storing plaintext passwords.
Although I get your point, this statement is a bit too broad. What about brute force? Get enough parallel hardware running fast enough and you can eventually reproduce the plaintext for any ciphertext or hashtext.
Besides brute force, there's also dictionary and rainbow table attacks to consider. Are you guilty if the plaintext can be reproduced by a table lookup? Are you guilty if you properly salted, but the salt storage was compromised too? Are you guilty if you enforced password-complexity requirements to foil a rainbow table, which led to the user writing down KuteK1tty!123 on her Post-it and a co-worker stole it?
These aren't easy questions, and there's no magic solutions, just a cloud of less-bad options.
Since we're on the subject of nitpicking, it's worth noting that under such circumstances you can produce a plaintext, and you can produce every plaintext of a certain length, but you cannot with certainty produce the plaintext unless there is only one possible plaintext meeting the constraints of the system.
Its not that black & white either. Many people still use MD5 which isnt reversible but still easily crackable, so where to draw the line ? Nobody would know if they were using MD5 but the risks would potentially be the same.
The point is, we do NOT know if FON is saving passwords in plain-text based on the provided info.
1. plain-text or plain-text equivalent: you can access the original passwords in microseconds
2. lousy hashing: you can't tell what the password is immediately but it's computationally feasible to figure it out
3. good hashing: you can't figure out the password
The only time you ever need to distinguish plaintext from plaintext-equivalent is when you have a partial data breach. Good hashing is safe even under a full data breach.
For all I care (and I guess the down-voters in your case think similar) that's just the same thing, really. If you can reproduce the plaintext in any way you're guilty of storing plaintext passwords. Arguing about the terms doesn't change the problem nor the perception in this humble author's opinion. The article is ~correct~, good enough, works for me.