You are viewing a plain text version of this content. The canonical link for it is here.
Posted to general@gump.apache.org by Stephen McConnell <mc...@apache.org> on 2004/05/14 23:22:09 UTC
gump stats looking odd
According to the following stats summary - one one gump project was
successfull. What is impled by the 512 "No Works" item?
-----------------------------------------------------------------------
Projects Successes Failures Prereqs No Works Packages
563 01 (0.18%) 00 (0.00%) 00 (0.00%) 512 (90.94%) 50 (8.88%)
-----------------------------------------------------------------------
Cheers, Steve.
--
|---------------------------------------|
| Magic by Merlin |
| Production by Avalon |
| |
| http://avalon.apache.org |
|---------------------------------------|
---------------------------------------------------------------------
To unsubscribe, e-mail: general-unsubscribe@gump.apache.org
For additional commands, e-mail: general-help@gump.apache.org
Re: gump stats looking odd
Posted by Stephen McConnell <mc...@apache.org>.
Adam R. B. Jack wrote:
>>>>Projects Successes Failures Prereqs No Works Packages
>>>>563 01 (0.18%) 00 (0.00%) 00 (0.00%) 512 (90.94%) 50 (8.88%)
>
>
>>Currently the success is showing as:
>>
>> Overall project success : 9.06%
>
>
> The 50 at the end are actually packaged *projects* (not simply modules in a
> Gump sense). See:
>
> http://brutus.apache.org/gump/public/packages.html#Packaged+Projects
>
>
>>This suggests that No Work entries may be factored into the calculation
>>of success.
>
>
> Depends upon if you consider a packaged project a "no work" or not. I chose
> to include it as work, because even thought it never shows as a success
> (Gump doesn't build it) it can count as a failure (if incorrectly
> installed/missing). As such 'overal success' is projects plus sucessfully
> installed packged projects.
OK - I understand - but I disagree. Packaging IMO should be factored
into an overhead of a real project build. That was - when I look at
gump info if its 50% then I know that half of Apache build successfully.
I.e. no micro economic adjustments required.
>>>From a total of 563 projects, 512 were counted as no-work -
>>leaving a total of 51 projects.
>
>
> Yup, the one done (to that point) plus those 50 packaged projects.
But the one project is not associated with the 50 packages (which I
think reinforces the notion that packaging is a overhead attributed to a
particular project).
>> 51/563 = 0.091
>
>
> Or 0.0906 (to give 9.06%).
Woops - yes.
>>But looking at the other numbers, 50 of these remaining projects are
>>modules. So in fact the the success ratio calculation is being skewed by
>>the inclusion of modules.
>
>
> Hey, we are cutting the Gumpmeisters some slack for correctly installing the
> package (and keeping them maintained).
:-)
> I'm not trying to skew, so I'll
> adjust if folks feel strongly, but this seems the right calculation to me.
Keep in mind that I'm they guy that assumed that an increase in fog was
a bad thing!
But aside from that - yep - something is not right here. IMO a failed
package due to a definition error deserves a gump message but it should
count in the overall (or partial) success factor. But take this just as
a comment from casual observation who has not contributing a single
character to the gump codebase!
> Thanks for looking closely at the numbers. :-)
No problem.
Cheers, Steve.
> regards,
>
> Adam
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: general-unsubscribe@gump.apache.org
> For additional commands, e-mail: general-help@gump.apache.org
>
>
--
|---------------------------------------|
| Magic by Merlin |
| Production by Avalon |
| |
| http://avalon.apache.org |
|---------------------------------------|
---------------------------------------------------------------------
To unsubscribe, e-mail: general-unsubscribe@gump.apache.org
For additional commands, e-mail: general-help@gump.apache.org
Re: gump stats looking odd
Posted by "Adam R. B. Jack" <aj...@trysybase.com>.
> >>Projects Successes Failures Prereqs No Works Packages
> >>563 01 (0.18%) 00 (0.00%) 00 (0.00%) 512 (90.94%) 50 (8.88%)
> Currently the success is showing as:
>
> Overall project success : 9.06%
The 50 at the end are actually packaged *projects* (not simply modules in a
Gump sense). See:
http://brutus.apache.org/gump/public/packages.html#Packaged+Projects
> This suggests that No Work entries may be factored into the calculation
> of success.
Depends upon if you consider a packaged project a "no work" or not. I chose
to include it as work, because even thought it never shows as a success
(Gump doesn't build it) it can count as a failure (if incorrectly
installed/missing). As such 'overal success' is projects plus sucessfully
installed packged projects.
> From a total of 563 projects, 512 were counted as no-work -
> leaving a total of 51 projects.
Yup, the one done (to that point) plus those 50 packaged projects.
> 51/563 = 0.091
Or 0.0906 (to give 9.06%).
> But looking at the other numbers, 50 of these remaining projects are
> modules. So in fact the the success ratio calculation is being skewed by
> the inclusion of modules.
Hey, we are cutting the Gumpmeisters some slack for correctly installing the
package (and keeping them maintained). I'm not trying to skew, so I'll
adjust if folks feel strongly, but this seems the right calculation to me.
Thanks for looking closely at the numbers. :-)
regards,
Adam
---------------------------------------------------------------------
To unsubscribe, e-mail: general-unsubscribe@gump.apache.org
For additional commands, e-mail: general-help@gump.apache.org
Re: gump stats looking odd
Posted by Stephen McConnell <mc...@apache.org>.
Adam R. B. Jack wrote:
>>According to the following stats summary - one one gump project was
>>successfull. What is impled by the 512 "No Works" item?
>
>
> "No Work" means -- "no work has been done on this entry (yet)". It used to
> show up when I'd do a partial run (e.g. depot-*), on all but those biult,
> but I've pretty much filtered those out.
>
>
>>-----------------------------------------------------------------------
>>Projects Successes Failures Prereqs No Works Packages
>>563 01 (0.18%) 00 (0.00%) 00 (0.00%) 512 (90.94%) 50 (8.88%)
>>-----------------------------------------------------------------------
>
>
> Had me going for a moment, or few, since clearly (looking at the build log)
> lots of projects are in other states.
>
> I see it is a side effect of (the new) re-building the buildLog.xml page
> after every module update, and after every project build, so it can (one day
> soon) be effectively a dynamic running log. This is to allow users to know a
> Gump is active, and where it is (in it's run).
>
> Unfortunately I went through a phase of worrying about walking down the tree
> (I had some expensive tree recursions), so got into a habit of doing things
> one-time & caching it on the object. Seems I tripped up on that here, and
> this froze after the first one was done.
>
> I've stopped making this one-time, and I'll check in shortly, so we'll see.
> Thanks for spotting this.
NP
Also related to this is the calculation of build success.
Currently the success is showing as:
Overall project success : 9.06%
This suggests that No Work entries may be factored into the calculation
of success. From a total of 563 projects, 512 were counted as no-work -
leaving a total of 51 projects.
51/563 = 0.091
But looking at the other numbers, 50 of these remaining projects are
modules. So in fact the the success ratio calculation is being skewed by
the inclusion of modules. In fact the correct success ratio should be:
(51-50)/(563-50)-512 == 1/513-512 == 1/1 = 100%
Does this make sense?
Cheers, Steve.
--
|---------------------------------------|
| Magic by Merlin |
| Production by Avalon |
| |
| http://avalon.apache.org |
|---------------------------------------|
---------------------------------------------------------------------
To unsubscribe, e-mail: general-unsubscribe@gump.apache.org
For additional commands, e-mail: general-help@gump.apache.org
Re: gump stats looking odd
Posted by "Adam R. B. Jack" <aj...@trysybase.com>.
> According to the following stats summary - one one gump project was
> successfull. What is impled by the 512 "No Works" item?
"No Work" means -- "no work has been done on this entry (yet)". It used to
show up when I'd do a partial run (e.g. depot-*), on all but those biult,
but I've pretty much filtered those out.
> -----------------------------------------------------------------------
> Projects Successes Failures Prereqs No Works Packages
> 563 01 (0.18%) 00 (0.00%) 00 (0.00%) 512 (90.94%) 50 (8.88%)
> -----------------------------------------------------------------------
Had me going for a moment, or few, since clearly (looking at the build log)
lots of projects are in other states.
I see it is a side effect of (the new) re-building the buildLog.xml page
after every module update, and after every project build, so it can (one day
soon) be effectively a dynamic running log. This is to allow users to know a
Gump is active, and where it is (in it's run).
Unfortunately I went through a phase of worrying about walking down the tree
(I had some expensive tree recursions), so got into a habit of doing things
one-time & caching it on the object. Seems I tripped up on that here, and
this froze after the first one was done.
I've stopped making this one-time, and I'll check in shortly, so we'll see.
Thanks for spotting this.
regards
Adam
---------------------------------------------------------------------
To unsubscribe, e-mail: general-unsubscribe@gump.apache.org
For additional commands, e-mail: general-help@gump.apache.org