Github user spmallette commented on the issue:
https://github.com/apache/tinkerpop/pull/892
> Long sounds fine.
I don't think that `Long` by itself is good enough. As I mentioned in my
comments, you can't assume you know anything about the identifier. TinkerPop
has no knowledge of what it contains. It could be that a `Graph`
implementation uses numbers that exceed the size of `Long` which would generate
an error in your current code.
> The sort algorithm I suggested is indeed a heuristic, still a better
heuristic as the default String sort by Java
What you have may be better for identifiers that have pure numerics, but
imo, it makes sorts of strings mixed with numbers less intuitive that the
default. I think that if you want your change for numbers then you need to make
them both make sense.
> let's go the other way and make debug/normalization easier accessible by
command line. I suggest adding the debug option even in the TinkerPop Docu
GraphML initial mentioning (I suggest to add this option for other formats as
JSON as well).
Your link in the above didn't work, but I think I know what you are
referring to. I guess I don't have a problem with calling more attention to the
normalization option, but I can't say that I fully follow why normalization
should be so important to anyone.
The only use case that I am aware of is if you were doing file
comparisons/diffs on generated files - normalization would help you there,
especially if you were pushing such files into source control or something
similar. But, to me, that's not a common use case. I would venture to say that
most people just want to "save their data" and don't really care about looking
at it in any way.
Do you have some other use case that makes this feature so important? You
mentioned "analysis of regression tests" but I would like to hear more about
that. We tend to be careful about features that don't suit wide use cases.
> Sorry for the import reordering,
No problem - happens all the time with all different IDEs
A final point, depending on the complexity of how this sorting algorithm
ends up, I think that it might be important for us to see what the performance
differences are between your changes and what we had before. That's nothing
that we need to have immediately as the PR is still changing, but I just wanted
to let you know that it may be necessary to fully evaluate this PR.
---
