[Pipet Devel] New guy speaks up

J.W. Bizzaro bizzaro at geoserve.net
Tue Nov 30 17:06:46 EST 1999


Brad Chapman wrote:
> 
> Hello brave Locians!

Hey new guy!

> In addition, I've been sorting through the
> old archives and reading up on all of the discussion I've missed

That's a huge task :-)

> and also
> trying to become more familiar with CORBA/GTK/bonobo/python. I meant to
> hang back silently a while longer and try and learn more, but I've been
> getting too excited about Loci and so I just felt the uncontrollable need
> to speak up with what I've been thinking about. I apologize if I am too out
> of touch still!

No problem.  It's nice to see someone new get excited about the project.

>         First let me give you a little background on myself so you can have
> an idea where I'm coming from. I'm a just starting off grad student in
> botany at the U of Georgia and although my background is in molecular
> biology/botany stuff, I've been directing my graduate thesis work towards
> including computational work--both out of necessity and interest (well,
> mostly interest!)

It would be unusual, and unusually nice, if your advisor were to let you take
on a computational project.  Most established biologists, chemists and
physicists, who _could_ get into bioinformatics, will not because (1) it is
just too different (for them and from their respective fields) and (2) they've
never used computers for anything beyond running shrink-wrapped applications.

> and have been trying to springboard off of my limited
> background in programming (C++) into the wide world of open source
> programming.

It isn't enough that we're on the bleeding edge being bioinformaticists, we've
got to be in the (currently) blazing hot field of open-source software!

> My interest in loci comes
> from the fact that my work will include integrating a number of previously
> existing programs and hopefully some distributed computing (so I don't have
> to run all of my sequence alignments on my pieced together home computer
> and wait 12 days for them to finish!).

Ah, you just described the goal of Loci in one sentence.

> In addition, I love the ideas behind
> Loci and want to see it succeed.

Oddly enough, so do I ;-)

>         So, because of all my interest I've been trying to get up to speed
> and so I have a few questions/discussion items from the depths of the
> mailing list archives:

Fire away.

> 1) First off, I really like the new core. I can connect dots together. Very
> nice--excellent work, Jeff! I do get some seriously weird behavior (Weird
> behavior 1) if I 1) add a document, 2) add a converter, 3) link them
> together (a 0 from the document to a 0 from the converter) I get some weird
> crazy stuff--a green 0/0 coming from the document which isn't connected to
> anything and a green 0/0 from the converter which moves together with a red
> 0 from the document.

Okay, I just tried what I think you tried.  And the problem is simply because
I didn't put any code in to handle this situation yet.  I think you tried
connecting an output line to another output line (or input to input).  Lines
with arrows that point away from the locus are outputs, and lines with no
arrows are inputs.  It's just something a user shouldn't be able to do, and
I'll make it so they simply won't connect. (If you want to know the dirty
details, in the code, inputs can only connect to outputs and visa versa.  If
you drop an output 0 on an output 0, Loci thinks you're dropping output 0 on
_input_ 0, and this leads to some crazy things in the code.)  Really though,
the fix is simpler than typing this paragraph :-)

> Weird Behavior 2) If I am really stupid and place dots
> from the same loci onto one point (ie. put a bunch of points from a
> processor onto one connection) I lose the connection and get a couple of
> huge green circles from the processor. Trippy!) These aren't complaints or
> anything (it's brand new--who could complain!?!), just what I've noticed!

Ooops.  Another no-no I need to handle.  Of course, the user shouldn't be able
to connect a locus directly back to itself.  In the workflow (when the script
is run), this would create an infinite loop.

(I'll answer the rest of your message in the next e-mail.)


Cheers.
Jeff
-- 
                      +----------------------------------+
                      |           J.W. Bizzaro           |
                      |                                  |
                      | http://bioinformatics.org/~jeff/ |
                      |                                  |
                      |           THE OPEN LAB           |
                      |    Open Source Bioinformatics    |
                      |                                  |
                      |    http://bioinformatics.org/    |
                      +----------------------------------+




More information about the Pipet-Devel mailing list