-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Peter,
Some questions inline:
On 02/16/2015 02:37 PM, Peter Palaga wrote:
(1) Tying to WildFly only. The change might be understood like
this. I am not sure this was your intention, but please consider
that importing another similar BoM (say EAP BoM) in paralel would
in fact be extremely impractical, because of version conflicts.
Aren't we targeting Wildfly? If so, then it makes sense to tie it to
Wildfly, at least for this bootstrap phase. Tying to a specific
application server at the beginning removes the need to solve problems
that we would have if we were targeting multiple application servers.
In other words, we can concentrate better on the business when we
don't need to make decisions about the integration/environment.
About EAP and similar derivatives: it should be close enough to
Wildfly to provide a mostly smooth transition. A possible solution
would be to use profiles, so that we import the appropriate BOMs
depending on the build type. But I'd suggest to worry about it once it
becomes a real problem.
Unless we know already that we *need* to also be able to run on other
application servers not compatible with Wildfly and that a refactoring
later is too expensive to be done at this later point.
(2) Pulling in many artifacts we do not use, do not need, some of
them maybe being implicitly unwanted ones. But having them in
parent is actually legalizing them without any discussion. That is
especially dangerous.
I might have missed your point here, but when you import a BOM, aren't
you just importing the dependency definitions? In such case, you can
always override on our parent if there's a unwanted dependency (with
an excludes). For the consumer of the dependency, it's usually easier
to just import a dependency without specifying the version, as there's
usually no reason why I would use a very specific version of a
library. And in such unusual cases, we can, again, override it on our
parent.
(3) "Political" dependence on WildFly. WildFly may do
changes at
any time in their BoMs that may have bad consequences in Hawkular.
Shouldn't matter if we stick with an already released version of
Wildfly and its BOM, right?
All in all, I completely agree with your points. I just do not think
it's the right time to be concerned about those points *yet*. All the
platforms we intend to run on do support Wildfly, which is "good
enough" for a first phase.
- - Juca.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1
iQEcBAEBAgAGBQJU4f1OAAoJECKM1e+fkPrXIIAIAIMvk2JnLVz0w9GPpUC+Hw0x
0UM+ADNDHQdNhs7GHlHhszbewAdfAOk2HBpq645M5Eg66t8HLIh7EW8/r9D1VD0r
okBLLs2WNf9z9HehKFiLuHSM6iziZ4njypGrOMQoX5B1ErqwmrrcHBnzn1pI8ayM
INoBsjVjrxNbL3m/CtNgw9+4yc8bhBTZI7cMt2Zud8JESFLlXmzshi0lB6udnT8X
ytfY0ZyRNpA/uBcVPrFw5lPP3xc0qbqLWRfRbRcXXHtMiTt6ybQQUWEPyrdmAROu
gGbHg6P6Ya/jmtd9fbkHGynGvVg0E3kEyWsnugIuEl5fXhAdHoASZy1y2qpLMVE=
=REZt
-----END PGP SIGNATURE-----