[CP2K:2064] Re: BerliOS bug tracker

[Teodoro Laino]

Ondrej,

Ondrej Marsalek wrote:
> Teo,
>
> thanks for the explanation, but I hope I understand what is expected
> of a proper bug report. The question was more
> technical/organizational.
>   
well.. I hope so.. although I'm a bit skeptical.. your old requests on 
this mailing list were never supported by input files..
pure speculation makes everything very difficult and slow (one should be 
extremely motivated to re-create the problem
you mention.. another thing is to have everything ready to go.)
I just found useful (for you as well) to remark this point.
> If BerliOS is broken, I would expect you (the developers) to want to
> move the project somewhere else. There is a bunch of good project
> hosting websites. It even looks like at least one of the developers
> was considering a move, perhaps? :-)
>
> http://code.google.com/p/cp2k/
>   
that was a pure trial setup done 2 years ago.. but we decided to stay 
with Berlios.
> Of course there are other options as well and I certainly do not want
> to start a flamewar about this, whatever works for you. A single
> comment though - I consider a proper (and working) bug tracker better
> than the current "official" bug reporting procedure, both for users
> and developers. But as I said, whatever works for you.
>   
Indeed, personally speaking, I find very convenient this way. It is very 
annoying to associate a certain bug
to a person. It may easily happen that one is overwhelmed with many 
reports.
Sending a message here:
1) all users are informed about a possible problem
2) the person that has some free time can decide to dedicate some time 
to it. If not at least there is a public
trace of the problem.

Moreover some of the bug reports done previously in Berlios were not 
really bug reports.. but more compilation/usage
issues. And is very difficult to train people how to use it..

At the present stage (we are almost 200 people subscribed here) the 
amount of requests is not so big that cannot
be handled (if correctly done) with  this simple (maybe pragmatic) way.

Regards,
Teo

> Cheers,
> Ondrej
>
>
> On Thu, May 7, 2009 at 10:44, Teodoro Laino <teodor... at gmail.com> wrote:
>   
>> Ondrej,
>>
>> BerliOS is not reliable (weird behavior and form time to time the
>> network collapses).
>> Last year we had an off time of almost 1 week!
>>
>> If you want to bug report something, create an archive with the input
>> file + all necessary things,
>> and post it here on the google mailing list + send a message explaining
>> things which are wrong..
>> Keep in mind that bug reports must run extremely quick, for many reasons:
>> 1) our time is limited
>> 2) easy things are easier to be debugged
>> 3) when creating easy  input files, in 50% of the times, people find
>> themselves the error they were doing..
>>
>> If the input files reflects the above points BE SURE that your request
>> will be listened.
>>
>> A very good way to be ignored is just archive an input file that is kind
>> of production (high cutoff.. hundreads of atoms..
>> large basis set..).. These are general rules, again, that people should
>> always keep in mind when asking for help.
>>
>> Just one example: let's say that you have a problem with REFTRAJ + CUBE
>> FILES. well.. you can build a very simple
>> bug report test suite using 1 water molecule + low cutoff + small basis
>> + ... that can run in few seconds..
>>
>> Or: if you have a problem with the integrator you can just post a test
>> suite using classical forcefield. This makes things 3 orders of
>> magnitudes faster.
>>
>> Regards,
>> Teo
>>
>> Ondrej Marsalek wrote:
>>     
>>> Dear CP2K developers,
>>>
>>> I would like to ask if there is any point in using the bug tracker at
>>> BerliOS [1] that is linked to from the CP2K homepage. I would be more
>>> than happy to use it and create proper bug reports, but both of the
>>> reports currently present in the tracker are somewhat discouraging.
>>>
>>> Best regards,
>>> Ondrej
>>>
>>> [1] http://developer.berlios.de/bugs/?group_id=129
>>>
>>>       
>>     
>
> >
>