[Biojava-dev] Potential Enhancements, Defect

Kalle Näslund kalle.naslund at genpat.uu.se
Tue Nov 1 07:50:24 EST 2005


Michael McCormick wrote:

> Greetings,
>
> Ruihan Wang and I are developing an application that uses biojava in  
> a J2EE environment. We have made a few changes and would like to add  
> them to the biojava code. All of the changes except for one class  
> involve serialization issues. Here is a brief summary.
>
> Please let me know if you are interested in adding these changes and  
> how they should be submitted.
>
> Thanks.
> Mike
>
> Michael McCormick
> Systems Analyst
> Fred Hutchinson Cancer Research Center
>
>
> /org/biojava/bio/search/SeqSimilaritySearchHit should be Serializable
> /org/biojava/bio/search/SeqSimilaritySearchResult should be Serializable
> /org/biojava/bio/search/SeqSimilaritySearchSubHit should be Serializable
> /org/biojava/bio/seq/FeatureHolder should be Serializable
> /org/biojava/bio/seq/db/SequenceDB should be Serializable
> /org/biojava/bio/symbol/Symbol should be Serializable
> /org/biojava/bio/symbol/SymbolList should be Serializable
>
> /

Hi!

Some of the Interfaces you are talking about are realy Core interfaces, 
such as SymbolList. It is used
in many different ways all trough biojava. Some implementations cant 
realy be serialized in a good way.
For example the BioSQL classes that are just proxies for the data inside 
the RDBM. So if you were to take
this type of SymbolList object, serialize it to disk, and load it back 
later, the RDBM might have changend,
and you have a very wierd situation. Similar situations exist for 
FeatureHolder and SequenceDB and so on.

Many of the actual implementations of FeatureHolder, SequenceDB and 
SymbolList and so on implement
Serializable though. So as long as you just use those, you can serialize 
FeatureHolders, SequenceDBs
just fine.

/kalle


More information about the biojava-dev mailing list