Well, we can't make a circle. We could have three:

jet, core collections (maps and lists), more math (vectors and matrixes)

but this is probably hopeless, and sticking with one is better.

On Mon, Dec 14, 2009 at 12:45 PM, Jake Mannix <[email protected]> wrote:
> But much of the math stuff is built out of the collections stuff (what is a
> vector
> but a collection of doubles and ints...)
>
> On Mon, Dec 14, 2009 at 9:42 AM, Benson Margulies 
> <[email protected]>wrote:
>
>> Would heads explode if I suggested two new components: mahout-math and
>> mahout-collections, the second depending on the first?
>>
>> On Mon, Dec 14, 2009 at 11:59 AM, Jake Mannix <[email protected]>
>> wrote:
>> > As far as I'm concerned, we can probably do this in steps, as long as we
>> > stabilize before 0.3, so no pushing down for now.
>> >
>> >  -jake
>> >
>> > On Mon, Dec 14, 2009 at 8:53 AM, Drew Farris <[email protected]>
>> wrote:
>> >
>> >> That's a good question. I was not going to push them down, but if
>> >> you/others believethat they should be pushed down at this point I'd be
>> >> happy to do so.
>> >>
>> >> On Mon, Dec 14, 2009 at 11:41 AM, Jake Mannix <[email protected]>
>> >> wrote:
>> >> > This all makes sense, I'm just wondering about the fact that now in
>> the
>> >> > top-level
>> >> > in o.a.m.math, we have Vector, Matrix, SparseVector, etc. from the
>> >> original
>> >> > Mahout
>> >> > linear algebra stuff.
>> >> >
>> >> > Should these not get pushed down into the math.matrix subdirectory
>> >> (except
>> >> > for
>> >> > the functions which go down into math.function)?
>> >> >
>> >> > I can see *not* pushing them down, because it completely intermingles
>> our
>> >> > old
>> >> > code with the Colt stuff, but at some point, they're going to be
>> >> > intermingled, right?
>> >> >
>> >> >  -jake
>> >> >
>> >> > On Mon, Dec 14, 2009 at 8:32 AM, Drew Farris <[email protected]>
>> >> wrote:
>> >> >
>> >> >> As far as the package structure is concerned, Benson proposed
>> >> >> something similar to the the following:
>> >> >>
>> >> >> o.a.m.matrix -> o.a.m.math
>> >> >> o.a.m.jet -> o.a.m.math.jet
>> >> >>
>> >> >> Things currently under o.a.m.matrix would retain their existing
>> names,
>> >> >> with s/matrix/math, for example:
>> >> >>
>> >> >> o.a.m.math.bitvector
>> >> >> o.a.m.math.buffer
>> >> >> o.a.m.math.function
>> >> >> o.a.m.math.list
>> >> >> o.a.m.math.map
>> >> >> o.a.m.math.matrix
>> >> >>
>> >> >> The maven submodule would be changed from mahout-matrix to
>> mahout-math
>> >> >>
>> >> >> As far as committing cleanup changes -- it is whatever is in the
>> patch
>> >> >> to MAHOUT-222, which I have not reviewed Once 222 is committed I'll
>> >> >> submit the reorg patch and more cleanup can proceed from there. I'm
>> >> >> certainly open to alternatives as well, just trying to jump in where
>> I
>> >> >> can help.
>> >> >>
>> >> >> Drew
>> >> >>
>> >> >> On Mon, Dec 14, 2009 at 11:23 AM, Jake Mannix <[email protected]
>> >
>> >> >> wrote:
>> >> >> > So we're committing cleanup changes first, and then doing reorg,
>> >> right?
>> >> >> >
>> >> >> > I'm for o.a.m.math, but we need to decide on where stuff goes in
>> >> >> subpackages
>> >> >> > below that, and then where to put stuff as far as maven submodules
>> >> too.
>> >> >> >
>> >> >> >  -jake
>> >> >> >
>> >> >> > On Mon, Dec 14, 2009 at 8:11 AM, Drew Farris <
>> [email protected]>
>> >> >> wrote:
>> >> >> >
>> >> >> >> No worries Sean -- it is really great to have your attention on
>> the
>> >> >> issue.
>> >> >> >>
>> >> >> >> On Mon, Dec 14, 2009 at 11:00 AM, Sean Owen <[email protected]>
>> >> wrote:
>> >> >> >> > No it's my bad, I missed one word while reading that message too
>> >> fast
>> >> >> >> > and misunderstood. Sorry for over-eagerly jumping in, I'll stay
>> out
>> >> of
>> >> >> >> > this.
>> >> >> >>
>> >> >> >
>> >> >>
>> >> >
>> >>
>> >
>>
>

Reply via email to