[Mono-dev] Bug Day

[Julien Gilli]

Paul F. Johnson wrote:

>> However, bringing this on the mono mailing lists  again is a good idea :-).
>>    
>>
>
>It was my son that reminded me - he thinks the idea of bugs in software
>is funny (he is 7 and to him, bugs are what you squish with a shoe!)
>
>  
>
:-).

>>>Are we going to have a Gnome style bug splatting day on the 3rd or 10th
>>>Sept?
>>> 
>>>      
>>>
>>Not on the 3rd, probably not on the 10th neither if nobody among key 
>>developers speaks up.
>>    
>>
>
>I think we should poke them with a stick. Not a nasty stick, possibly
>one with a feather on the end ;-)
>
>  
>
Does any developer read this thread ? If so, could you please tell us 
how relevant our idea is ?

>>Maybe we should setup the first bug day without wating from comments and 
>>see if everything goes well.
>>    
>>
>
>A dry run? Sounds like it could be interesting.
>
>  
>
A dry run, indeed.

>>The only problem is that many of us won't have the needed credentials to 
>>modify bugs status, so we'd have to e-mail the developers and tell them 
>>"we think that this bug should be marked as this, we were able to 
>>reproduce it on the following platform, under these conditions, etc.".
>>    
>>
>
>Which is the idea of the triage (from my understanding). Bugs roll in,
>tested and if it fails due to not a programming fault, then it gets
>passed to the code doctors who know it's a genuine problem and not just
>someone messed up.
>  
>
I totally agree on that.

>As to having the credentials, you're right. In the past I've been part
>of a porting effort to get gcc onto RISC OS and parts of that were mind
>blowing! From what I've seen, mono is almost as bad, but does have the
>advantage of C# being quite an easy language which is simple enough to
>follow.
>  
>
By credentials, i meant that any non developer can't modify a bugs 
status, IIRC. This is annoying if we want to triage bugs, since we have 
to ask someone who has the right to modify bug status to do so.

>What I would suggest is this. If we can get one of the key developers
>(Jordi, Ben, Peter, Miguel or anyone else for that matter) and a couple
>of those who understand the compilers innards (patch submitters) and do
>a couple of dry run scenarios (say we take a random selection of bug
>reports), we try and get the faults to occur and then pass them over.
>  
>
and we tell them how we would modify the bug properties (branch 
concerned, milestone, component, priority, severity, etc.)

>These faults don't have to be real ones either. 
>
>  
>
Can you tell us more about this ? I don't understand how we could use 
fake bugs.

>This will serve to do a couple of things. Firstly, it will help with
>time difference problems and any communication difficulties. It'll also
>demonstrate the work required plus give us (the non Novell people) a
>work out!
>  
>
Again, I totally agree with you.

All the best,

-- 
Julien Gilli
IDEALX http://www.idealx.com/