> BUT, this could result in having tens of libs ~= the same. Eg:
> turbine-3.0-20020527,
> turbine-3.0-20020101, turbine-3.0-20010203, etc... This would fill up
> the lib.repo at
> light's speed, and would make it really confusing, never knowing which
> one to use.
> This problem grows exponentially with the number of dependencies.
> 
> IMO, the single lib repo approach forces the developper to choose
> between one of both
> approaches : stable release OR cvs snapshot.
> 
> If you have enough developpers to work on all your projects in the
same
> time, and if you
> want to use an evolving code, then you choose the "-dev", so that all
> coders adapt in the
> same time to the same lib, keeping easy the integration (two projects
> don't rely on
> different APIs).

Hm, good points. For the most part agree with you; -dev should be
preferred over timestamps in most situations.

What motivated my post was that I spent a long time trying to figure out
an error that Torque 3.0-b2 was giving me. The more I think about it,
the less convicted I am that -dev conflicts the problem I was seeing,
but regardless Torque 3.0-b2 has 5 of it's 13 lib files as -dev
versions.

So my case was probably rare enough not to worry about; I had a released
application that relied on -dev version where those -dev versions where
in the lib.repo (my doing) and could be over-written with newer ones.

Also, as I browse the Jakarta.apache.org/turbins/jars repository, many
of the supposedly -dev versions are old. There are a lot from January,
February, and March. If the point of -dev is to be running with the
latest, greatest code, shouldn't these be updated on a fairly frequent
basis?

Thanks,
Stephen



--
To unsubscribe, e-mail:   <mailto:[EMAIL PROTECTED]>
For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>

Reply via email to