On 6/2/2011 5:58 AM, Andrew Bennetts wrote:
> Public bug reported:
>
> Try this:
>
> $ cd /tmp
> $ bzr branch lp:ubiquity
>
> It quickly estimates about 70k records. When it reaches that mark it
> keeps going, regularly bumping the estimate up a few thousand as it does
> so. In the end it reaches over 150k, but for over half of that time the
> estimate gives the impression that it's almost done.
>
> You could almost consider this to be two bugs: that the initial estimate
> is off by a factor of 2, and that having exceeded that estimate you then
> have no idea how much longer it will really take because it continually
> suggests it's nearly done.
>
> ** Affects: bzr
> Importance: Medium
> Status: Confirmed
>
>
> ** Tags: ui
>
We've talked about changing the fetch api to be able to give hints to
this effect. If we switched fetch to work in batches of revisions and
their associated content, we could also estimate #revisions much more
accurately. (Rather than just 'total number of records', which depends
on how wide the tree is (for chk pages), and how many files are changed.)
John
=:->
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (Cygwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
On 6/2/2011 5:58 AM, Andrew Bennetts wrote:
> Public bug reported:
>
> Try this:
>
> $ cd /tmp
> $ bzr branch lp:ubiquity
>
> It quickly estimates about 70k records. When it reaches that mark it
> keeps going, regularly bumping the estimate up a few thousand as it does
> so. In the end it reaches over 150k, but for over half of that time the
> estimate gives the impression that it's almost done.
>
> You could almost consider this to be two bugs: that the initial estimate
> is off by a factor of 2, and that having exceeded that estimate you then
> have no idea how much longer it will really take because it continually
> suggests it's nearly done.
>
> ** Affects: bzr
> Importance: Medium
> Status: Confirmed
>
>
> ** Tags: ui
>
We've talked about changing the fetch api to be able to give hints to
this effect. If we switched fetch to work in batches of revisions and
their associated content, we could also estimate #revisions much more
accurately. (Rather than just 'total number of records', which depends
on how wide the tree is (for chk pages), and how many files are changed.)
John
=:->
-----BEGIN PGP SIGNATURE----- enigmail. mozdev. org/
s56oACgkQJdeBCY SNAANsnwCbBHMtb z6gelJlIzZeHlij /6bS ii0kMINb2EQ/ 3g6Mk
Version: GnuPG v1.4.9 (Cygwin)
Comment: Using GnuPG with Mozilla - http://
iEYEARECAAYFAk3
fncAnRGLwDCpnID
=lp6G
-----END PGP SIGNATURE-----