1 |
On Mon, 2007-05-14 at 14:25 -0700, Daniel Ostrow wrote: |
2 |
> |
3 |
> We release our packages as upstream intends. If they don't split them, |
4 |
> we don't split them, talk to upstream not us. |
5 |
|
6 |
Well that is not always the case. Not to contradict. |
7 |
|
8 |
We tend to make packages in the Java world that are subsets of a |
9 |
upstream package. In short it's because many upstreams bundle |
10 |
dependencies. So we will strip out those dependencies and maintain them |
11 |
as separate packages. Even though it was not really intended to be such |
12 |
by upstream. |
13 |
|
14 |
For example, take dev-java/servletapi ( really tomcat-servlet-api same |
15 |
sources, soon to be merged with virtual ... ). Many packages ship with a |
16 |
servlet-api.jar. That jar comes from Tomcat, but is not available as |
17 |
package on it's own. It's sources were in a separate CVS tree with its |
18 |
own build system, but as of Tomcat 6. That was merged into unified |
19 |
source tree and single build system. |
20 |
|
21 |
Upstream releases a project called Glassfish, that will eventually be |
22 |
many glassfish-* packages. Although due to the size Glassfish is broken |
23 |
up but there are not individual sources available per sub-project. JBoss |
24 |
is very similar, and many packages use JBoss modules. Like |
25 |
glassfish-jsf-api :) Mighty fun stuff. |
26 |
|
27 |
We do try to get with upstream were it's feasible. But really much if |
28 |
this is out of control in the Java world. I believe Maven is an attempt |
29 |
to reign in some level of control, manageability, and etc. Among other |
30 |
things. |
31 |
|
32 |
Granted all this might be unique to Java. Since applications developed |
33 |
in other languages, don't always have a binary release available. When |
34 |
one is, rarely do they ship all the libraries in binary format they |
35 |
need. |
36 |
|
37 |
But packages do get split up even though upstream does not really |
38 |
provide for such. I would not recommend doing it unless there is a |
39 |
reason and need like on the Java front. |
40 |
|
41 |
Sorry for length. |
42 |
|
43 |
-- |
44 |
William L. Thomson Jr. |
45 |
Gentoo/Java |