Gábor Csárdi

I was thinking whether we should put the error class definitions in a separate package. You might define `assertionError`, but other packages might want to use this error type in...

YEah, something like that. I think I would make it even more bare, and not assume anything about how the conditions are thrown. Just define them, `lookup_error = condition(...)`, and...

That's a good point, yes, custom messages should be allowed and encouraged. But we can still avoid signalling the condition. So I think what you have now is quite good....

Sure, that makes a lot sense. So #23 essentially means a packaging mechanism, which would be just great to have! Imho packaging is where Python is not so strong. This...

I think the module manager can be a module. For easier installation you can also build native installers for it e.g. a DEB package for ubuntu, a DMG for OSX,...

No, I mean, is "running" modules from the shell already solved? Did you have something similar to Python's `if __name__ == "__main__":` idiom, or am I misremembering?

I suspect that this will be hard/impossible, if you want to support using the same class names in multiple modules. Just a heads-up.

> S4 classes have an attribute that specifies the associated package, don’t they? Could this be hijacked for modules? Indeed a good idea, and basically our only hope. :) But...

This is a bit more difficult, as we need to model the read-length distribution and the likelihood function will be a bit more difficult.

Hi, can you share a (preferably) small GFF and SAM/BAM file that reproduces the error? Thanks. On Tue, Apr 30, 2013 at 10:23 AM, rahulcdeo [email protected]: > Hi, > >...