[Bioperl-l] Re: Bio::DB::WebSeqDB

[Hilmar Lapp]

Aaron J Mackey wrote:
> 
> > Jason Stajich wrote:
> 
> > > b) Write our own LWP::Protocol class which extends LWP::Protocol::http
> > >    and processes the stream as it goes rather than reading the whole thing
> > >    in, not sure about the feasibility here.
> 
> While this would seem to be the most desirable case, in actuality I'm not
> so sure: You'd be asking the http connection to remain open for as long as
> you needed it.  Now if all you were doing was converting formats, no big
> deal, but if you were piping each sequence to some analysis program, I'm
> not sure you could count on the web server on the other side of the http
> connection to keep the connection open.  I could be wrong, though, as I'm
> definitely not a web server expert.  I know many servers have KeepAlive
> turned off (usually used for server-push applications), not sure how this
> would impact this application.
> 

The performance penalty concern I had was that downloading the sequences
may actually take the same time as processing them. So, if you could
start firing something in the background on your first seq while the
second keeps loading, it may save you significant time on a slow network
connection, and the server certainly wouldn't shut down (would it?) the
connection in the middle of the download. But again, I don't think this
is even a close to average use case of Bioperl, so it is probably
overkill to try to accommodate this situation efficiently. People,
correct me if I'm wrong.

	Hilmar
-- 
-----------------------------------------------------------------
Hilmar Lapp                                email: hlapp@gmx.net
GNF, San Diego, Ca. 92122                  phone: +1 858 812 1757
-----------------------------------------------------------------