commons-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Claude Warren <cla...@xenei.com>
Subject Re: Discussion: New commons module/project
Date Wed, 14 Feb 2018 09:27:37 GMT
@Charles,
I thought of Commons because it is where people go to find common code.  it
provides a web presence to base the code and discussion from.  Though
perhaps Github would suffice.

@Ralph,
I agree that reference implementations would be good.  But I didn't want to
start a "language war" with respect to what languages to support.  I think
that if there is a discussion and a consensus reached we are 90% of the way
to reference implementations.  What may be more critical are reference
tests.

I will come back to my discussion of Bloom filters because there is so much
variance in how they are implemented in the wild.  Bloom filters are used
in a number of products and (as far as I know) there is no place to go for
a clear, concise description of how to create them.  In fact the Cassandra
code has a comment block that indicates their modified code method is
faster and produces the same range of results as other methods.

Another algorithm that I can not find any detail on is the blockchain
mining algorithm.  How is a minor notifiied that a new block has been added
to the chain and who do they notify when they add a new block.  There are
lots of high level descriptions but no low level algorithm type info that I
have been able to find.

For algorithms like merge-sort or b-tree there are lots of books and other
publications that describe the algorithms, but many of the "newer"
constructs do not.  I see this project as a place developers can turn to to
find the information they need to implement various algorithms, and if not
to start a discussion about how to correctly implement it.

Claude







On Wed, Feb 14, 2018 at 3:33 AM, Charles Honton <chas@honton.org> wrote:

> What can / does commons provide that cannot be done with GitHub?
>
> chas
>
> > On Feb 13, 2018, at 5:58 PM, Ralph Goers <ralph.goers@dslextreme.com>
> wrote:
> >
> > If this was a project to create specs AND provide reference
> implementations I think it would make sense. I don’t see how a project that
> just creates specs fits with Commons personally.
> >
> > Ralph
> >
> >> On Feb 13, 2018, at 1:46 PM, Matt Sicker <boards@gmail.com> wrote:
> >>
> >> A sort of commons-algorithms type library? That kind of strikes me as
> what
> >> Commons is in the first place. I could see it being broken down into
> >> components, though. For example, commons-graph <
> >> https://commons.apache.org/sandbox/commons-graph/> (seems inactive) for
> >> graph algorithms, commons-tree for trees, heaps, and other similar ADTs
> and
> >> algorithms. Some ADTs already belong in commons-collections. Some
> >> algorithms may already be in commons-math, and there's efforts toward
> >> splitting that up into individual components (e.g., rng, numbers,
> >> statistics), so it doesn't make sense to group them in there IMO.
> >>
> >> On 13 February 2018 at 05:23, Bruno P. Kinoshita <
> >> brunodepaulak@yahoo.com.br.invalid> wrote:
> >>
> >>> Hi Claude,
> >>>
> >>> Quite sure there was similar discussion some time ago. But I can't
> recall
> >>> if it was here in commons, somewhere in the incubator, or labs?
> >>>
> >>> But regarding commons, before the component/jira/mailing list are
> created,
> >>> I think it would have to go either via sandbox or incubator first?
> >>>
> >>>
> >>> Perhaps having some code somewhere like GitHub to show more or less the
> >>> concept would make it easier for others to evaluate the idea?
> >>> Cheers
> >>> Bruno
> >>>
> >>>
> >>>
> >>>
> >>>
> >>> ________________________________
> >>> From: Claude Warren <claude@xenei.com>
> >>> To: Commons Developers List <dev@commons.apache.org>
> >>> Sent: Tuesday, 13 February 2018 11:55 PM
> >>> Subject: Re: Discussion: New commons module/project
> >>>
> >>>
> >>>
> >>> The goal is arrive at a common understanding of the specific algorithm
> and
> >>> potentially one or more code examples for implementation which may be
> >>> pseudo code.
> >>>
> >>> I think all this requires is:
> >>>
> >>>  1.  jira (or similar) to track the discussions and mark them when they
> >>>  are closed
> >>>  2. A mechanism to list the algorithms that have been or are under
> >>>  discussion.  Perhaps the Jira search could provide this by default.
> >>>  3. A repository for code snippets.  Though again perhaps Jira would be
> >>>  sufficient.
> >>>  4. Mailing list in order to take votes and the like.
> >>>
> >>> Claude
> >>>
> >>>
> >>>
> >>> On Tue, Feb 13, 2018 at 10:31 AM, Bernd Eckenfels <
> ecki@zusammenkunft.net>
> >>> wrote:
> >>>
> >>>> Hello,
> >>>>
> >>>> I am not sure how this might turn out, is it only discussion or do you
> >>>> also want to develop a specification language/toolset? In either case
> it
> >>>> does not sound like the typical commons sub-project. What
> infrastructure
> >>>> would you require?
> >>>>
> >>>> Gruss
> >>>> Bernd
> >>>> --
> >>>> http://bernd.eckenfels.net
> >>>> ________________________________
> >>>> From: Claude Warren <claude@xenei.com>
> >>>> Sent: Tuesday, February 13, 2018 10:36:55 AM
> >>>> To: Commons Developers List
> >>>> Subject: Discussion: New commons module/project
> >>>>
> >>>> Disclaimer:  I am not certain of the proper nomenclature to use here
> so
> >>>> when I say module consider that it may be a sub-project or project
> under
> >>>> commons.  The term should be taken to mean a collection of the things
> >>> under
> >>>> discussion.
> >>>>
> >>>> I would like to propose a commons module for the development and
> >>>> documentation of algorithms.  As anyone who has tried can attest,
> writing
> >>>> specifications that are not misunderstood is nigh on impossible.  This
> >>>> module would provide an area to discuss various algorithms and work
> out
> >>>> code examples in various languages for implementation.
> >>>>
> >>>> For example: Bloom Filters.  When creating a bloom filter you take the
> >>>> modulus of a digest to determine which bit in a bit vector to turn on.
> >>> But
> >>>> the modulus of a negative number is negative.  So is it proper to take
> >>> the
> >>>> absolute value (abs)  of the modulus or the digest or are you
> required to
> >>>> consider the digest to be a very long unsigned integer?  All of these
> >>>> choices have repercussions and may impact interoperability.
> >>>>
> >>>> I have several co-workers and friends that are interested in exploring
> >>>> algorithms of this sort.  Is there any interest here in starting such
> >>>> discussions?  If so how does one go about it?
> >>>>
> >>>> I would like to see a ticketing system (Jira?) that discussions could
> be
> >>>> conducted against.  So, for example, a single ticket for bloom filter
> >>>> implementation.  The ticket would stay open until consensus was
> achieved.
> >>>> Once achieved the result would be documented and references to the
> >>>> discussion provided.
> >>>>
> >>>> Not clear on where to document the results and whether or not that
> >>>> documentation would include code.
> >>>>
> >>>> Thoughts?
> >>>> Claude
> >>>>
> >>>>
> >>>> --
> >>>> I like: Like Like - The likeliest place on the web
> >>>> <http://like-like.xenei.com>
> >>>> LinkedIn: http://www.linkedin.com/in/claudewarren
> >>>
> >>>>
> >>>
> >>>
> >>>
> >>> --
> >>> I like: Like Like - The likeliest place on the web
> >>> <http://like-like.xenei.com>
> >>> LinkedIn: http://www.linkedin.com/in/claudewarren
> >>>
> >>> ---------------------------------------------------------------------
> >>> To unsubscribe, e-mail: dev-unsubscribe@commons.apache.org
> >>> For additional commands, e-mail: dev-help@commons.apache.org
> >>>
> >>>
> >>
> >>
> >> --
> >> Matt Sicker <boards@gmail.com>
> >
> >
> >
> > ---------------------------------------------------------------------
> > To unsubscribe, e-mail: dev-unsubscribe@commons.apache.org
> > For additional commands, e-mail: dev-help@commons.apache.org
> >
> >
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: dev-unsubscribe@commons.apache.org
> For additional commands, e-mail: dev-help@commons.apache.org
>
>


-- 
I like: Like Like - The likeliest place on the web
<http://like-like.xenei.com>
LinkedIn: http://www.linkedin.com/in/claudewarren

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message