You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@harmony.apache.org by Stepan Mishura <st...@gmail.com> on 2008/04/16 09:24:36 UTC
[testing] How long we should keep test results for millstones?
Hi folks,
We have accumulated test results for several milestones: M2-M5 [1][2][3][4]?
I think that it doesn't make sense to keep them forever.
Should we archive them or remove entirely? Are they still useful?
Sure, M5 tests results will be used for M6 regression tracking.
But still we should make a decision what to do with old results.
[1] http://people.apache.org/~mloenko/snapshot_testing/script/r551077/index.html
[2] http://people.apache.org/~mloenko/snapshot_testing/script/r580985/index.html
[3] http://people.apache.org/~mloenko/snapshot_testing/script/r603534/index.html
[4] http://people.apache.org/~mloenko/snapshot_testing/script/r629320/index.html
Thanks,
Stepan
Re: [testing] How long we should keep test results for millstones?
Posted by Stepan Mishura <st...@gmail.com>.
On 4/19/08, Endre Stølsvik <En...@stolsvik.com> wrote:
> Stepan Mishura wrote:
> >
> > > Kinda compromise is possible, to compress older data and provide a
> > > link to the bulk archive if someone needs detais. Summary pages do not
> > > eat much anyway.
> > >
> > >
> >
> > OK. I'll archive tests results for M2-M4 and keep the archives for a
> while.
> >
>
> Can't you move'em over to Harmony Proper Webspace? With a link to a small
> super-simple "performance tests" page, linking out to the 2 "current" tests,
> and the archives of the older?
> The "hiding away" of these interesting stuffs on private space shouldn't be
> necessary in the first place, IMHO.
>
I'm not going to keep them privately. Currently archives are available
in my home folder [1][2][3]. May be it make sense to move them to
released milestones binaries on the archive site [4].
[1] http://people.apache.org/~smishura/harmony-test-results-M2-r551077.tgz
[2] http://people.apache.org/~smishura/harmony-test-results-M3-r580985.tgz
[3] http://people.apache.org/~smishura/harmony-test-results-M4-r603534.tgz
[4] http://archive.apache.org/dist/harmony/milestones
Thanks,
Stepan.
> Endre.
>
Re: [testing] How long we should keep test results for millstones?
Posted by Endre Stølsvik <En...@stolsvik.com>.
Stepan Mishura wrote:
>
>> Kinda compromise is possible, to compress older data and provide a
>> link to the bulk archive if someone needs detais. Summary pages do not
>> eat much anyway.
>>
>
> OK. I'll archive tests results for M2-M4 and keep the archives for a while.
Can't you move'em over to Harmony Proper Webspace? With a link to a
small super-simple "performance tests" page, linking out to the 2
"current" tests, and the archives of the older?
The "hiding away" of these interesting stuffs on private space
shouldn't be necessary in the first place, IMHO.
Endre.
Re: [testing] How long we should keep test results for millstones?
Posted by Stepan Mishura <st...@gmail.com>.
On 4/16/08, Alexey Varlamov <al...@gmail.com> wrote:
<SNIP>>
> AFAIR number of files was the main issue, though space may become an
> issue too in some future :)
>
The space became an issue - I've just got a notification from Apache
Intra team that the size of my home dir is too big.
> Kinda compromise is possible, to compress older data and provide a
> link to the bulk archive if someone needs detais. Summary pages do not
> eat much anyway.
>
OK. I'll archive tests results for M2-M4 and keep the archives for a while.
Thanks,
Stepan.
> --
> Alexey
>
> >
> > Regards,
> > Tim
> >
>
Re: [testing] How long we should keep test results for millstones?
Posted by Alexey Varlamov <al...@gmail.com>.
2008/4/16, Tim Ellison <t....@gmail.com>:
> Endre Stølsvik wrote:
> > Tim Ellison wrote:
> >
> > > Alexey Petrenko wrote:
> > >
> > > > I do not think that we are really need any other test results then
> Mn-1
> > > >
> > >
> > > I agree, keep the last Milestone for regression checking, and if you
> really want to be nostalgic keep the last two, but anything beyond that is
> unnecessary.
> > >
> >
> > Why not keep everything? I guess the space it takes isn't quite the issue?
> >
> > You can more easily see trends in the data with more than two datapoints,
> uncovering more subtle regressions etc. And it may even have a tint of "fun"
> to it, seeing how much better it became over time, for each milestone,
> revision and version.
> >
> > (Of course, these data can be recreated by checking out the older
> milestones and run the tests again... but ..)
> >
>
> If space is not an issue then yes, we might as well keep them around, but we
> share finite resources with the rest of Apache so I think we should only
> keep what we think we will need.
>
> If somebody has an idea to use it for archaeological purposes then fine.
AFAIR number of files was the main issue, though space may become an
issue too in some future :)
Kinda compromise is possible, to compress older data and provide a
link to the bulk archive if someone needs detais. Summary pages do not
eat much anyway.
--
Alexey
>
> Regards,
> Tim
>
Re: [testing] How long we should keep test results for millstones?
Posted by Tim Ellison <t....@gmail.com>.
Endre Stølsvik wrote:
> Tim Ellison wrote:
>> Alexey Petrenko wrote:
>>> I do not think that we are really need any other test results then Mn-1
>>
>> I agree, keep the last Milestone for regression checking, and if you
>> really want to be nostalgic keep the last two, but anything beyond
>> that is unnecessary.
>
> Why not keep everything? I guess the space it takes isn't quite the issue?
>
> You can more easily see trends in the data with more than two
> datapoints, uncovering more subtle regressions etc. And it may even have
> a tint of "fun" to it, seeing how much better it became over time, for
> each milestone, revision and version.
>
> (Of course, these data can be recreated by checking out the older
> milestones and run the tests again... but ..)
If space is not an issue then yes, we might as well keep them around,
but we share finite resources with the rest of Apache so I think we
should only keep what we think we will need.
If somebody has an idea to use it for archaeological purposes then fine.
Regards,
Tim
Re: [testing] How long we should keep test results for millstones?
Posted by Endre Stølsvik <En...@stolsvik.com>.
Tim Ellison wrote:
> Alexey Petrenko wrote:
>> I do not think that we are really need any other test results then Mn-1
>
> I agree, keep the last Milestone for regression checking, and if you
> really want to be nostalgic keep the last two, but anything beyond that
> is unnecessary.
Why not keep everything? I guess the space it takes isn't quite the issue?
You can more easily see trends in the data with more than two
datapoints, uncovering more subtle regressions etc. And it may even have
a tint of "fun" to it, seeing how much better it became over time, for
each milestone, revision and version.
(Of course, these data can be recreated by checking out the older
milestones and run the tests again... but ..)
Endre.
Re: [testing] How long we should keep test results for millstones?
Posted by Tim Ellison <t....@gmail.com>.
Alexey Petrenko wrote:
> I do not think that we are really need any other test results then Mn-1
I agree, keep the last Milestone for regression checking, and if you
really want to be nostalgic keep the last two, but anything beyond that
is unnecessary.
Regards,
Tim
> 2008/4/16, Stepan Mishura <st...@gmail.com>:
>> Hi folks,
>>
>> We have accumulated test results for several milestones: M2-M5 [1][2][3][4]?
>> I think that it doesn't make sense to keep them forever.
>>
>> Should we archive them or remove entirely? Are they still useful?
>>
>> Sure, M5 tests results will be used for M6 regression tracking.
>> But still we should make a decision what to do with old results.
>>
>> [1] http://people.apache.org/~mloenko/snapshot_testing/script/r551077/index.html
>> [2] http://people.apache.org/~mloenko/snapshot_testing/script/r580985/index.html
>> [3] http://people.apache.org/~mloenko/snapshot_testing/script/r603534/index.html
>> [4] http://people.apache.org/~mloenko/snapshot_testing/script/r629320/index.html
>>
>> Thanks,
>> Stepan
>>
>
Re: [testing] How long we should keep test results for millstones?
Posted by Alexey Petrenko <al...@gmail.com>.
I do not think that we are really need any other test results then Mn-1
2008/4/16, Stepan Mishura <st...@gmail.com>:
> Hi folks,
>
> We have accumulated test results for several milestones: M2-M5 [1][2][3][4]?
> I think that it doesn't make sense to keep them forever.
>
> Should we archive them or remove entirely? Are they still useful?
>
> Sure, M5 tests results will be used for M6 regression tracking.
> But still we should make a decision what to do with old results.
>
> [1] http://people.apache.org/~mloenko/snapshot_testing/script/r551077/index.html
> [2] http://people.apache.org/~mloenko/snapshot_testing/script/r580985/index.html
> [3] http://people.apache.org/~mloenko/snapshot_testing/script/r603534/index.html
> [4] http://people.apache.org/~mloenko/snapshot_testing/script/r629320/index.html
>
> Thanks,
> Stepan
>