Re: [Xen-devel] Prototype Code Review Dashboards (input required)

[Lars Kurth <lars.kurth.xen@xxxxxxxxx>]

> On 9 Mar 2016, at 17:06, Daniel Izquierdo <dizquierdo@xxxxxxxxxxxx> wrote:
> 
> On 02/03/16 23:45, Daniel Izquierdo wrote:
> 
> [...]
>>> Given, that the Xen-A1.A2.A3 dash board is quite busy, we should keep that 
>>> separate.
>>> 
>>> 
>>> 
>> 
>> I still have to upload this, sorry for the delay!
>> 
>> 
>> This is a summary of the actions unless extra comments are provided:
>> 
>> * Balance should be calculated as reviews - patches
>> * Cleaning actions in the dataset when finding multiple email addresses
>> * Bugs with the post-ack comments
>> * Add the extra panel defined as use case A.0
>> 
>> Some extra feedback or discussion:
>> 
>> * Reduce the flags to be used. We're currently using all of the flags 
>> available and sub-setting the reviewed-by and acked-by
>> * I'm not sure if we need extra discussion related to the merge:1 filter
>> * Check how reviews and patches are counted. Any new version of a patch is 
>> counted as a new patch. Any new reviewed-by flag in a reply to a patch is 
>> counted as a review.
> 

Thank you for the progress update

> From those actions, this is the current status:
> 
> * Balance should be calculated as reviews - patches (DONE)
> * Cleaning actions in the dataset when finding multiple email addresses 
> (still in progress and I should add your comments in a previous email)
> * Bugs with the post-ack comments (DONE)
> * Add the extra panel defined as use case A.0 (DONE)
I noticed that the naming of fields in the panels have inconsistent names, e.g. 
patchserie_sender_domain vs sender_domain
That makes writing queries hard and, if possible, should be cleaned up at some 
stage

> * Tables at the bottom of the panels contains proper information and not all 
> of the fields (DONE)
> * Do not count comments to your own patches (to be done)
Note that we thus should also not count post-ACK comments on own patches

> Regarding to the extra feedback:
> * Reduce the flags to be used. We're currently using all of the flags 
> available and sub-setting the reviewed-by and acked-by.
> ** This opened a new discussion about the CC mailing lists and the creation 
> of some buckets of info.
We should discuss the best way forward

> * I'm not sure if we need extra discussion related to the merge:1 filter
> ** No advances
Are there any areas where you need my input. I think there were 2 issues
a) One generally with uninitialised values : however that applies to other 
flags also
b) One about the views not shown : I think that is OK and fixing a) would make 
it clear what values are defined and which are not  

> * Check how reviews and patches are counted. Any new version of a patch is 
> counted as a new patch. Any new reviewed-by flag in a reply to a patch is 
> counted as a review.
> ** No advances

I think these are both OK. Rationale:
- Reviewed by flags can be removed and then added again. As we count multiple 
review comments, we can count multiple reviewed-by
- As for counting a new version of a patch as a separate one, I think we need 
some discussion
Firstly, we need to be careful of is that we don't include old versions of a 
patches/patchseries within the uncommitted backlog.
Also, we need to make sure that counting of patch related items (e.g. reviews) 
are consistent with how we count patches.

Regards
Lars 

 

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel