[ODE] On the meaning of ODE stability (was some quick notes on 0; 6) ...

[Terry L. Triplett]

On 6/22/06, Rodrigo Hernandez <kwizatz at aeongames.com> wrote:
>
> Terry L. Triplett wrote:
>
> > Who are these mysterious others?  Will they not reveal themselves?   :-)
>
> You alone seem to be  the vocal minority who advocates for a system wide
> library,


Grrr.  No, I"m *NOT* advocating a system-wide library in the sense that you
imply.  I want a viable mechanism for *shared* libraries with clean
versioning, which is a different thing.

I'm probably in the minority because as a maintainer of a library that in
turn supports downstream users, I have to speak in proxy as their
advocate.   They are depending on *me* to work things out.  That's part of
the responsibility I accepted when I stepped up to maintain a project rather
than just take, take, take.  And because I'm working in an area not
currently part of the greater collective consciousness, the community is
small.  Throw me a bone here, for god's sake!!  :-)

I'm in the unenviable position of supporting a binding to ODE from a higher
level language (C# via P/Invoke) where static binding at compile time is
*not an option*.  Since binding happens at runtime, a shared library is the
only reasonable option.  The shared library needs to have a clearly
identifiable version so that in the context of the wider system environment,
the correct library is chosen, should an ambiguity exist.

You're confusing "System Wide Library" with "making a library available to
users via established mechanisms of distribution and following established
conventions".   A typical Linux distribution is capable of supporting
multiple library versions.  Applications and packages that depend on a
particular version of a library can depend on them.  But where does the
information about library versions come from?  I maintain that this is the
responsibility of the library creators/maintainers.  That's where the
questions/discussion about sonames came from, not because I want to
establish ODE as a "System Wide Library".

I remember John Watte supporting the
> position that a System Wide library was a bad idea the second to last
> time we discused sonames, (forgive me John if I am mistaken).


An soname is useful/important for specifying changes to the API/ABI,
regardless of whether a library is "System Wide" or not (and remember, a
"System Wide Library" can exist in multiple versions).   It is still
useful/important for upstream users to be able to distinguish between
library versions - why not use the established, in-place mechanisms.  How
else are the ODE developers keeping track of API/ABI changes?  Right now, it
seems to me that it's via a seat-of-the-pants, "what changed since the last
release?  Do you know?  No, not me." mechanism.

> Targeting features rather than labels is a better path for this kind
> > of project, IMHO.
>
> True, I am not going to argue with you here, as this seems to be arguing
> just for the sake of arguing.


No, it's trying to establish a sane foundation based on established
principles of project management.  Arbitrary labels don't work well.
Defined goals do.  That's not an argument for the sake of arguing, it's what
any experienced project manager learns on the job.  Who in the real world
works on the basis of arbitrary labels as opposed to specific requirements
and goals?

Look, all of this talk isn't intended to slam the ODE maintainers for the
great job they are doing, it should be considered as feedback from the ODE
community from folks that are trying to get something done.  Being branded
as a vocal minority isn't exactly motivation for delivering ODE to a
community of users that would otherwise not use this great library.

(In my darker moments, I think that a wrapper to ODE is a foolish
undertaking.  A 'native' C# physics library makes much more sense in the
long term.  Any takers?  I'm willing to give it a go ...)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://q12.org/pipermail/ode/attachments/20060622/77a21796/attachment.htm