Bitmag 1142 integrity gui display checksum age

[ole-v-v]

https://sbforge.org/jira/browse/BITMAG-1142
On the integrity web GUI, for each pillar display configured max age of checksums and age of currently oldest checksum.

[Bohlski]

Very nice.

Apart from my requested changes, I think it would be nicer looking if in the integrity-GUI the "Age of oldest checksum"-column showed maybe just the current largest unit of time and just had its lowest shown unit set to hours or something like that. Currently the seconds-count keeps updating on every call to /getIntegrityStatus, which is largely irrelevant when it could take years before we update checksums.

[ole-v-v]

Apart from my requested changes, I think it would be nicer looking if in the integrity-GUI the "Age of oldest checksum"-column showed maybe just the current largest unit of time and just had its lowest shown unit set to hours or something like that.

Thanks, I agree that currently it’s absurdly precise. On the other hand, should the value (unlikely) come down to a few days, I think that we do want hours too. Suggestion for a compromise: Include hours if months are 6 or less. Include minutes if days are 8 or less. Include seconds if days are 0. Never include fraction of second. This generally gives the user a precision of 0.5 % of the age or finer. I don’t find that too fine-grained. And when not perfectly sure, I’d still rather be too precise.

[m-atlantis]

Apart from my requested changes, I think it would be nicer looking if in the integrity-GUI the "Age of oldest checksum"-column showed maybe just the current largest unit of time and just had its lowest shown unit set to hours or something like that.

Thanks, I agree that currently it’s absurdly precise. On the other hand, should the value (unlikely) come down to a few days, I think that we do want hours too. Suggestion for a compromise: Include hours if months are 6 or less. Include minutes if days are 8 or less. Include seconds if days are 0. Never include fraction of second. This generally gives the user a precision of 0.5 % of the age or finer. I don’t find that too fine-grained. And when not perfectly sure, I’d still rather be too precise.

I think that we don't want the timer to update as often as it currently does, so showing seconds is kinda irrelevant - especially since whoever is looking don't really care if it's 45 seconds old.
I think it is gonna be used to see if something is 1 month old compared to years perhaps. And once we no longer need to care about seconds, then we definitely don't want to update the age very often

[ole-v-v]

The validation is in the wrong place here. It should be with the max age setting and should be documented there (since in practice the value comes from there). I suggest we fix that separately after both BITMAG-992 and BITMAG-1142 have been merged. We may also want to consider once more whether 500 years is an appropriate limit (java.time.Duration can handle over 292 000 000 000 years but java.time.Period only 2 100 000 000 years; we need a limit somewhere).