Re: ts.check() and ways of getting unresolved dependencies from a ts

Paul Nasrat <pauln-xGvoirzvwgBWk0Htik3J/[email protected]> Mon, 11 Oct 2004 16:20:56 +0000
Newsgroups gmane.linux.redhat.rpm.python
Message-ID <[email protected]>
On Mon, Oct 11, 2004 at 11:21:33AM -0400, Sean Dilda wrote:
> On Mon, 2004-10-11 at 11:05, seth vidal wrote:
> 
> > We could also have ts.check() return a list of 'tsproblem' objects.
> > 
> > These objects could have attributes like:
> > problem.type - sense
> > problem.need - needname/version/whatever
> > problem.whatneeds - header object of the package that is 
> >                     needed/conflicted, what have you
> > 
> 
> For the most part, I like this idea.  Especially with special objects,
> as opposed to just dictionaries (this is for cosmetic as well as
> backwards compatibility reasons).

I think list of objects is the way to go.

> This reminds me of how the output of os.stat() was changed in the core
> python libraries.  Originally it returned a tuple.  Not it returns an
> object that acts much like you're describing.  But they also made it
> backwards-compatible by making the new object indexable in the same way
> the tuple was.  I suggest following their lead and have it return a list
> of these new backwards-compatible objects.

That sounds really nice. I'll look into implementation  of a pseudotuple. 

PyStructSequence seems the key - will fiddle.

Paul