[
https://issues.apache.org/jira/browse/YETUS-144?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15169941#comment-15169941
]
Allen Wittenauer edited comment on YETUS-144 at 2/26/16 10:09 PM:
------------------------------------------------------------------
bq. suffer from an copy and paste bug.
Damn it! I thought I fixed that! Good catch!
bq. Isn't the function always there ?
Nope. If I tell test-patch to not use system plug-ins or I provide my own
maven plug-in or ... , there is no guarantee that function will exist. This is
one of the few (only?) of tests calling into something that doesn't come from
the main body of code. (The JIRA/github bits come to mind too, but that's
fairly specialized vs. this meaning to be generic). I suppose we could create a
dummy one in core.d somewhere and then let the maven plug-in file override it
during file load. That would guarantee existence and eliminate the existence
check.
Another thought I had was to add $\{BUILDTOOL\}_add_install as a standard
function, but this is likely just a maven-ism and plug-ins will still need some
knowledge that they need to use it in particular cases.
I admit that this whole scenario feels awkward as hell, but I'm not really sure
if there is any way around it to make it even remotely clean. :( I'm certainly
open to ideas! haha
was (Author: aw):
bq. suffer from an copy and paste bug.
Damn it! I thought I fixed that! Good catch!
bq. Isn't the function always there ?
Nope. If I tell test-patch to not use system plug-ins or I provide my own
maven plug-in or ... , there is no guarantee that function will exist. I
suppose we could create a dummy one in core.d somewhere and then let the maven
plug-in file override it during file load. That would guarantee existence and
eliminate the existence check.
Another thought I had was to add ${BUILDTOOL}_add_install as a standard
function, but this is likely just a maven-ism and plug-ins will still need some
knowledge that they need to use it in particular cases.
I admit that this whole scenario feels awkward as hell, but I'm not really sure
if there is any way around it to make it even remotely clean. :( I'm certainly
open to ideas! haha
> mvn install usually required on mvn builds
> ------------------------------------------
>
> Key: YETUS-144
> URL: https://issues.apache.org/jira/browse/YETUS-144
> Project: Yetus
> Issue Type: Bug
> Components: Test Patch
> Reporter: Allen Wittenauer
> Assignee: Allen Wittenauer
> Priority: Blocker
> Fix For: 0.2.0
>
> Attachments: YETUS-144.00.patch, YETUS-144.01.patch,
> YETUS-144.02.patch
>
>
> Found an interesting side effect of YETUS-137 this morning. Since the
> compile step was now skipped, the mvn install was also skipped. As a result,
> mvn test broke if the mvn cache was new.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
