From: Richard Maine on
TaliesinSoft <taliesinsoft(a)me.com> wrote:

> On 2010-05-18 02:35:15 -0500, Richard Maine said:
>
> > TaliesinSoft <taliesinsoft(a)me.com> wrote:
> >
> >> I will have 25 consecutive hourly backups ending with the most recent, and
> >> not the 24 that the Apple documentation states.
> >
> > As long as one is being that picky (and I'm not at all sure why you
> > care, but I'll play along for one post at any rate), one should quote
> > the documentation accurately. If you misquote it, then you don't get any
> > credit towards the game of picking nits.
>
> The first sentence in my opening posting in this thread was as follows:
[elided]
> I think I fairly and correctly posted that sentence, but I'm certainly
> open to having misinterpreted it.

Yes, I agree. But then in the post I responded to you referred to "the
24 [hourly backups] that the Apple documentation states." That is what I
was referring to as a misquote; the Apple documentation does not state
that. That part is your "translation". I would say that your translation
is in error. Yes, you initially quoted it correctly, but then you
switched to the "translated" version and still referred to it as what
"Apple documentation states," which makes it a misquote. That's what I
was complaining about.

> My interpretation has been that each backup covers an hour in time so
> that 24 backups would be needed to cover 24 hours.

I thought that was what you were thinking, but that is wrong, as I
tried, apparently unsucessfully, to explain. A backup does not "cover" a
time period. It is a backup at a single time point. There is absolutely
nothing in a backup to indicate anything about a time period.Note, for
example, that there is *NO* difference between an hourly and a daily
backup. The system doesn't do some kind of special backup and call it a
daily one. All it does is regularly delete the hourly ones. The ones
that it doesn't delete are then the daily ones. It doesn't change
anything about those backups to make them daily ones; it just doesn't
delete them.

If you have 24 hourly backups right after making one, then the oldest
one is 23 hours old - not 24. If you want to recover from an error made
23.5 hours ago, you are out of luck. There is no sense in which this
"covers" more than 23 hours.

This is a pretty common concept - nothing unique to Time Machine, or
even backups. It is much more fundamental concept of data organization.
A single sample does not "cover" anything; it is just a single time
point. In order to "cover" a range you need multiple samples, and the
range covered is that between the samples.

--
Richard Maine | Good judgment comes from experience;
email: last name at domain . net | experience comes from bad judgment.
domain: summertriangle | -- Mark Twain
From: TaliesinSoft on
On 2010-05-18 10:27:19 -0500, Richard Maine said:

[responding to my having stated]

>> My interpretation has been that each backup covers an hour in time so
>> that 24 backups would be needed to cover 24 hours.
>
> I thought that was what you were thinking, but that is wrong, as I
> tried, apparently unsucessfully, to explain. A backup does not "cover" a
> time period. It is a backup at a single time point. There is absolutely
> nothing in a backup to indicate anything about a time period.Note, for
> example, that there is *NO* difference between an hourly and a daily
> backup. The system doesn't do some kind of special backup and call it a
> daily one. All it does is regularly delete the hourly ones. The ones
> that it doesn't delete are then the daily ones. It doesn't change
> anything about those backups to make them daily ones; it just doesn't
> delete them.

Agreed that a backup doesn't cover a period of time but captures the
state of the system at a moment in time, however each backup "includes"
and "excludes" the differences, that is the changes, the deletions and
inclusions, that have occurred between it and the immediately preceding
backup. When one is performing hourly backups in sequence each such
backup in effect describes what has happened within the hour preceding
the time it was made.

Which takes us back to my original pondering as to why Time Machine
needs 25 consecutive hourly backups to identify the changes that have
occurred within the last 24 consecutive hours?


James Leo Ryan
Austin, Texas

From: TaliesinSoft on
On 2010-05-18 10:19:40 -0500, Wes Groleau said:

[in response to my having stated in response to a remark he made]

>> I do find it somewhat amusing that the "so much fuss" issue is raised
>> here when in other threads there are often literally hundreds of
>> "clever" comments which often have no relation to the originally posted
>> topic.
>
> If you read those thread (and I hope you don't!) you might notice that
> once in a great while I post some similar (and equally futile) attempt
> to wind it down.

I tend to avoid those threads, but will make an occasional check to see
if perhaps something of relevance has been posted.

--
James Leo Ryan
Austin, Texas

From: Richard Maine on
TaliesinSoft <taliesinsoft(a)me.com> wrote:

> On 2010-05-18 10:27:19 -0500, Richard Maine said:

> Agreed that a backup doesn't cover a period of time but captures the
> state of the system at a moment in time, however each backup "includes"
> and "excludes" the differences, that is the changes, the deletions and
> inclusions, that have occurred between it and the immediately preceding
> backup. When one is performing hourly backups in sequence each such
> backup in effect describes what has happened within the hour preceding
> the time it was made.

Not if there is no previous hourly backup, it doesn't. You keep trying
to introduce this concept of a single backup covering a time period.
Changing the terminology to talk about differences or anything else
won't help. It just doesn't do that. Only the period between backups is
covered. A backup does not record changes. (Well, a traditional
incremental backup would, but that's not what Time Machine does; that
would be *VERY* different).

If you have 24 hourly backups, you have no record of what changes
happened in the hour before the first one. Your record goes back to that
first one and then stops. If you try to say that it "in effect describes
what happened within the hour preceding when the backup was made", you
could just as well say that it described the same thing for a year
instead of just an hour. Both time periodds are equally fictitious.

If you have 24 hourly backups right after having made one, then you
can't recover from errors made 23.5 hours ago.

> Which takes us back to my original pondering as to why Time Machine
> needs 25 consecutive hourly backups to identify the changes that have
> occurred within the last 24 consecutive hours?

I thought I explained this, but I guess I failed to get the point
across. This isn't just Time Machine. It is the nature of any data
organization that includes point samples over a range. Having spent 40
years of my career working with that general kind of data, it is pretty
much second nature to me. I couldn't even point to where I learned it or
what I learned it from, though I'm sure I wasn't born with the notion.
Maybe it has become so "obvious" to me that I can't adequately explain
it to someone who doesn't "just get it". It isn't exactly a complicated
point, but I don't seem able to get it across.

I hereby give up.

--
Richard Maine | Good judgment comes from experience;
email: last name at domain . net | experience comes from bad judgment.
domain: summertriangle | -- Mark Twain
From: TaliesinSoft on
On 2010-05-18 11:05:45 -0500, Lewis said:

> It's been explained why 25 backups are required for 24 hours of backups.
> It's simple math.

As I write this it is a bit after 11 AM and say I'm running hourly
backups on the hour. The most recent backup covers 10 AM to 11 AM, the
next most recent 9 AM to 10 AM, and so on. In that case, if we continue
on backwards for a total of 24 backups we will come to 11 AM to 12 PM
on the preceding day, meaning that those 24 backups run from 11 AM on
one day to 11 AM on the next day, and that is 24 hours of backups. I'm
still at a loss as to why that 25th backup is needed.

--
James Leo Ryan
Austin, Texas