[Biophp-dev] I'm back and itching...

S Clark biophp-dev@bioinformatics.org
Mon, 27 Oct 2003 11:31:06 -0700

Somewhat distractedly - I've just become fully re-employed, and I've driven
2800 miles over the last two weeks (on-site visits - thankfully this will
normally only be once per month - the rest of the time I can telecommute).
I just got back last night, so sorry about my quiet-ness lately.

As previously reported, I have working frontends for NCBI's online BLAST
as well as local BLAST, and both use the same XML-based parser.  What remains
from my perspective is to add "intuitive" interface methods for setting
the parameters, which is actually not difficult at all, it just requires
that I get an hour or two uninterrupted to get them added.  The idea is
that these interface methods will have the same names between the local 
and web-based frontends, so that the use of them both will be as much 
identical as is possible.  The hard part lately is just getting the 
uninterrupted time...If anyone wants to look at them before I get around
to adding these features, I can always post a tarball of them as they are now.

Both ESummary and ESearch are in the same state, that is, "no-frills" but
both functional.  EFetch, on the other hand, will need separate parsers for
every individual record type (sequence, pubmed, etc.)

I have long agreed with Jesus' observation that we should probably NOT
take the approach of trying to make "BioPERL(/Java/Python) but written in 
PHP", though we should probably start thinking about how we DO want
to lay out the structure of the modules.  Should we split up the
sections by task ("IO,Conversion,Networking,Program Execution,etc.") or
by Object ("Sequence, Publication, Phylogenetic Trees, etc.")?  Or some
other scheme?

(Re-doing BioXXXX in PHP would be DOUBLY wasteful - not only because we'd
just be replicating what already exists, but also because of PHP's increasing
ability to integrate with Java, we could just as easily link directly to
BioJAVA modules...)

Indeed, I think this may be the solution to the 'slow' parts of PHP - making
use of Java modules for 'computationally intensive' processes.  With PHP5 
rapidly approaching what seems to be near-"native" integration with Java, 
this seems like a natural solution.

As a side note - one OTHER thing I've been working on as best I can is trying
to get my shiny new "Business Class" DSL line working properly so that I can
more easily host files and such directly.  It's possible I could 'donate' some
space and bandwidth to hosting some things if I can get this DSL service to do
what I paid for it to do...

Welcome back!

On Sunday 19 October 2003 01:17 am, Serge Gregorio wrote:
> Well, that's from my end.  Sean, how's your work going along?
> Serge