1 |
-----BEGIN PGP SIGNED MESSAGE----- |
2 |
Hash: SHA1 |
3 |
|
4 |
Greetings, |
5 |
|
6 |
Currently, 'dojar' is a script part of each sys-apps/portage release. I am |
7 |
proposing to deprecate all Java ebuild functions from Portage and move |
8 |
them into the Java eclass. |
9 |
|
10 |
In terms of what this means: |
11 |
|
12 |
1. It's easier to implement new functions in an eclass. |
13 |
2. We can roll out new functions faster instead of waiting for them to get |
14 |
checked into Portage and go through Portage's qa cycle. |
15 |
3. dojar would be immediately enhanced to allow more control over how |
16 |
CLASSPATH gets set in package.env. |
17 |
|
18 |
I would take initial responsibility for bumping all ebuilds that currently |
19 |
use 'dojar' to use the new eclass when it's ready. Currently only (6) |
20 |
ebuilds outside of dev-java use 'dojar': portagemaster, postgresql, |
21 |
jboss, orion, tomcat, pdflib. |
22 |
|
23 |
dojar would be replaced with java_dojar(). |
24 |
|
25 |
My initial plans for enhancing jar file installation will be for |
26 |
correctness in /usr/share/${PN}/package.env. I will be giving ebuilds the |
27 |
chance to prepend and/or append to the CLASSPATH setting. (Order |
28 |
sometimes matters.) |
29 |
|
30 |
For example, see bug #10360. To solve this bug, I would add the following |
31 |
to the ant ebuild: |
32 |
|
33 |
JAVA_DEP_PREPEND="junit" |
34 |
|
35 |
Multiple packages (space separated) could be supplied here. The packages |
36 |
can be whatever the developer wants, providing that the package(s) |
37 |
indicated exist and have a package.env. Additionally, there is |
38 |
JAVA_DEP_APPEND if one wants some dependencies at the _end_ of the |
39 |
CLASSPATH. In this case, it doesn't matter whether junit gets prepended |
40 |
or appended, as long as it ends up in the CLASSPATH. |
41 |
|
42 |
What happens next? In the example of #10360, the java eclass would in turn |
43 |
process that variable by reading junit's package.env, get the CLASSPATH |
44 |
and prepend that to the CLASSPATH it is writing for ant's package.env. |
45 |
|
46 |
This would close the bug and would also solve a lot of other similar bugs |
47 |
down the line. |
48 |
|
49 |
Comments? Flames? Agreements? |
50 |
|
51 |
Cheers, |
52 |
Dylan Carlson |
53 |
|
54 |
Public Key: http://pgp.mit.edu:11371/pks/lookup?op=get&search=0x708E165F |
55 |
Key fingerprint = 3AEA DE38 FE42 15A6 C0E2 730E 3D04 BCC1 708E 165F |
56 |
-----BEGIN PGP SIGNATURE----- |
57 |
Version: GnuPG v1.2.1 (GNU/Linux) |
58 |
|
59 |
iD8DBQE+pv3fPQS8wXCOFl8RAgxeAJ9RvDhRbNJiKhKWrMfoCk+edhecBwCdGKWx |
60 |
7AoM4bO9yj1Jq2m6QLXBC90= |
61 |
=cybU |
62 |
-----END PGP SIGNATURE----- |
63 |
|
64 |
|
65 |
-- |
66 |
gentoo-dev@g.o mailing list |