Hello !
> I have used it, although I wasn't using the fact it returned a Partition
> instance.
Oh. So what you needed was actually map(len, G.connected_components()).
Well, this functon has been added in order to return a partition (this the
name _to_partition), and this is one thing I find no good explanation for.
> I'd be okay with that (although it is somewhat long).
It is somewhat long and uncomfortable. I would prefer to remove it myself,
but if this is the only middle ground then let it be.
>> I do not understand what you mean by 'similar'. I tried to explain in my
>> first post how the behaviour of this function was not clear from its
name,
>> and how renaming it would make it almost a copy of a function that
already
>> exist. Also, if you want the graph corresponding to the hasse diagram,
it is
>> perfectly natural to do Graph(P.hasse_diagram()).
>
> You're saying remove the method because you can replace it with a short
> 1-line statement with that last sentence.
While it can indeed be rewritten with a short 1-line, and while I gave this
1-line repeatedly, please do not reduce my whole argumentation to that. The
reason why I believe that this function should be removed is that I find
its name unclear, as I explained in my first post on the subject.
In order to fix this problem, I believe that the name this function could
be renamed to should contain both ('undirected' or 'Graph') as well as
Hasse Diagram. This would give something like
Poset.undirected_hasse_diagram; or Poset.hasse_diagram_undirected; or
Poset.hasse_diagram_graph, none of which offers any simplification with
respect to the already available alternatives, i.e.:
- Graph(Poset.hasse_diagram())
- Poset.hasse_diagram().to_undirected()
Because it needs to be renamed, and because all names I can think of offer
no advantage over already available alternatives, I believe that this
function should be removed.
> It depends on how you think of posets. I don't naturally think of them as
> directed graphs, so I wouldn't a priori know that I could plug it into
> Graph.
This is the key point: there is indeed no unique Graph associated to a
Poset, which is probably the reason why the two of them are not 'naturally
associated' for you. Indeed, I oppose both Graph(Poset) and
Poset.to_graph(), as equally unclear.
While there is no 'uniquely determined' poset associated to a Poset, there
is a notion of Hasse Diagram, which is directed. To this directed graph we
can naturally associate an undirected graph, and this I believe is the best
way to obtain the desired result. It is faithfully represented by the two
commands I mentionned.
> I think we can only hope for so much with the names themselves as certain
> words are often overloaded from different areas of math and we don't want
> 40+ character function names.
We are not as desperate as you say. Besides, the problem with to_graph has
nothing to do with long names.
> Again, if you can think of a better, go ahead
> and change it.
I do not have one. I still defend that Poset.to_graph should be removed for
the reasons I gave above. Do you find them unfair on some respect ?
I am eager to see this matter solved.
Nathann
--
You received this message because you are subscribed to the Google Groups
"sage-devel" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to sage-devel+unsubscr...@googlegroups.com.
To post to this group, send email to sage-devel@googlegroups.com.
Visit this group at http://groups.google.com/group/sage-devel.
For more options, visit https://groups.google.com/d/optout.
