Estimating NNN/MMM

[Maritza Mendez]

Right.  It is not our imagination.  I almost always see an improper
fraction.  True, I have imported some massive (by bzr standards
apparently) trees with deep trees and large dynamic range in filesize.
 So I can imagine that the estimate could be poor but that it would be
updated as it gets smarter.


On 10/19/10, Talden <talden at gmail.com> wrote:
>>> So when I see something like Estimate M/N where M grows to equal and
>>> then exceed N, sometimes by more than a 2:1 margin, does that mean that
>>> the initial estimate was not especially good?
>>
>>> Thanks
>>> ~M
>>
>>
>> It shouldn't ever exceed N, as N should grow larger if M grows past it.
>> However, yes. N is not fixed, because it is only an estimate, which can
>> be updated.
>
> I'm pretty sure it's not our imagination... it's more usual for me to
> see M larger than N than the reverse (I had assumed it was not meant
> as a vulgar fraction of work done since it would typically be a value
> greater than 1).
>
>> Specifically, it is a fairly good estimate if you are working with a
>> long lived branch, but not for an initial import of a large branch
>> (little history, lots of file content).
>
> Ahh perhaps the cases in which I've noticed this are poor estimation cases.
>
> --
> Talden
>