[
https://issues.jboss.org/browse/JBIDE-18876?page=com.atlassian.jira.plugi...
]
Nick Boldt commented on JBIDE-18876:
------------------------------------
6 months later, no better suggestions for a workflow that would avoid the problem of
"waiting for bits to appear before pushing a new .target file to nexus".
maybe we could implement a sleep period between steps 4 and 5?
or we could composite "previous instance of the same snapshot version" with
"latest instance of the same snapshot version" into a single URL, as we do for
JBT component sites. That'll double the disk usage (each TP instance is 977 - 1.1G
because it includes both the TP and the TP zip), but might save these once-a-month
instances of TP updates causing builds to fail for 30-60 mins while a new change is
rebuilt.
Improve TP publishing so changes released to git or deployed to nexus
would not break developer local builds
------------------------------------------------------------------------------------------------------------
Key: JBIDE-18876
URL:
https://issues.jboss.org/browse/JBIDE-18876
Project: Tools (JBoss Tools)
Issue Type: Bug
Components: build
Affects Versions: 4.3.0.Alpha1
Reporter: Denis Golovin
Assignee: Denis Golovin
Fix For: 4.4.0.Alpha1
Latest Thym update is 'good' example for this problem. I was in the middle of
testing some changes in parent/pom.xml and suddenly build start to fail with Thym
resolution problem. IMO what happened is TP .target files were published to nexus before
actual p2-repos appeared online. Building from latest revision didn't help ether
because of the same problem. I had to revert to previous revision to continue my task.
So it would be good if TP builds publish binaries first and then release TP sources to
git and deploy to nexus.
Please note, that when sftp/rsync for unified TP binaries is finished it doesn't mean
p2-repos are available from
download.jboss.org right away.
--
This message was sent by Atlassian JIRA
(v6.4.11#64026)