[personal] Re: [MOBY-dev] [RFC] Exception Reporting in bioMOBYv1.5- INB proposal

[David González Pisano]

Ok, I admit it was not a good example ;-)

Martin Senger escribió:

>>corresponding DBMS. The input to the service is a Simple representing a 
>>protein, while the output is a Collection with all possible interactions 
>>where that protein is participating.
>>
>  This is a good example because it nicely illustrates what Mark said,
>and not a need for naming Simples in a Collection (IMHO) :-) :
>
>   1) First, you cannot have a service that can return *all* protein
>interactions. Such service is a sci-fi. So you go for "all possible"
>protein interactions. Well, but when a DBMS source is not available, its
>interactions are "not possible".
>  
>
Obviously. Maybe Jose María could explain it better, but we are not 
talking about the biology here, and probably the example should say *all 
possible protein interactions stored in my databases", ok. But the topic 
is why should I want to report an error for a Simple in a Collection, 
not if is possible to obtain all interactions for a given protein or not ;-)

The example (like all examples) should help us to extrapolate and 
imagine real use cases. In this case, I can imagine another service 
where I want to annotate an input probe list collection (those present 
in my microarray), for example against several databases with different 
information. Some of them are not available at execution time, or cannot 
find the required information because the query does not accept a given 
input character, so I want to inform the user that I have a warning for 
this specific probe, because I was not able to retrieve the information 
from the db due to whatever problem had happened.

>   2) Also, you do not know what interactions *would* be found in the
>non-responding DBMS, so you cannot create any Simple representing such
>non-returned interactions.
>  
>
Again, let's suppose my Simple needs the name *and* the type and 
description of the interaction, and that pieces of information are 
stored in different DB's. What if I can retrieve the name (ie, not an 
empty simple) but cannot retrieve its type, description, etc... (ie, my 
object is not empty but cannot be completed...)?

>   3) And last, you can still inform the users about not covering all my
>resources (this time) in a general warning in mobyException - but there is
>really no sense to put this warniong in a collection itself.
>  
>
I still think that the general case (a whole resource needed for the 
service or the collection is not available) is different than the 
particular case (a single error, warning or information could be 
generated for a particular Simple inside an input Collection, while the 
other Simples are all right).

>   I must admit that I have not paid much attention to this topic when it
>was explained in Malage. Because now, I really do not see why we need it.
>
Whatever. Is an optional part in the proposal, and depends on a BioMOBY 
specification change to be possible to implement, so there is no problem 
in removing it from the proposal if the community thinks is not 
necessary to specify it. We will save the file for future use if needed ;-)

David
-------------- next part --------------
A non-text attachment was scrubbed...
Name: dgpisano.vcf
Type: text/x-vcard
Size: 338 bytes
Desc: not available
URL: <http://lists.open-bio.org/pipermail/moby-dev/attachments/20050902/8f080999/attachment.vcf>