singa-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "ASF subversion and git services (Jira)" <j...@apache.org>
Subject [jira] [Commented] (SINGA-488) Travis CI always build from Apache master branch
Date Fri, 27 Sep 2019 03:25:00 GMT

    [ https://issues.apache.org/jira/browse/SINGA-488?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16939088#comment-16939088
] 

ASF subversion and git services commented on SINGA-488:
-------------------------------------------------------

Commit b1475fc722a780e6fa1aecb3b2bb197377e9686d in incubator-singa's branch refs/heads/master
from Chris Yeung
[ https://gitbox.apache.org/repos/asf?p=incubator-singa.git;h=b1475fc ]

SINGA-488 Change the path of source code for CI

> Travis CI always build from Apache master branch
> ------------------------------------------------
>
>                 Key: SINGA-488
>                 URL: https://issues.apache.org/jira/browse/SINGA-488
>             Project: Singa
>          Issue Type: Bug
>            Reporter: Moaz Reyad
>            Priority: Blocker
>          Time Spent: 20m
>  Remaining Estimate: 0h
>
> It seems that the Travis build always use the master branch from github.com/apache/incubator-singa.git
instead of building the pull request.
> When checking the job log of Travis, I found it reads from github two times:
> 1. In the first time, it reads the correct pull request: (for example Pull Request 532)
> {code:bash}
> git clone --depth=50 https://github.com/apache/incubator-singa.git apache/incubator-singa
> ...
> git fetch origin +refs/pull/532/merge:
> git checkout -qf FETCH_HEAD
> {code}
> 2. Then it ignores this correctly fetched code, and does another git clone from the conda
build script:
> {code:bash}
> Cloning into bare repository '/home/travis/conda-bld-1594.2/git_cache/github.com/apache/incubator-singa.git'...
> ...
> Your branch is up to date with 'origin/master'.
> {code}
> This means that it always builds the master branch instead of building the pull request.
So it has two problems:
>  # Missed alarms: if the master branch builds successfully, this will be shown as if
the pull request is fine and can be merged. While the pull request code was actually never
built.
>  # False alarms: if the master branch does not build successfully, this will be shown
as if the pull request has a problem, even if the problem is actually in the master branch
not in the pull request.
> If this issue is correct, then this is a blocker issue and I suggest to resolve it before
creating or merging any pull request. Because the continuous integration tools that is supposed
to test the pull requests is not working correctly.



--
This message was sent by Atlassian Jira
(v8.3.4#803005)

Mime
View raw message