[ 
https://issues.apache.org/jira/browse/LUCENE-7024?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15144214#comment-15144214
 ] 

Uwe Schindler commented on LUCENE-7024:
---------------------------------------

bq. The script crawls pom.xml.template files from the release branch of the 
source repo (as a source of truth about what Maven artifacts should be 
produced), and crawls all of the maven artifacts, POMs included, from the RC 
URL, which is either an http(s):// URL (now served by the ASF Subversion 
server), or is local (a file:// URL).

Why does it need to fetch it from Subversion or GIT? I think the most correct 
would be to take them from the .src.tgz.gz file that was downloaded previously 
and extracted. This would also be an "authoritative" source. The smoker does 
also not download the full source code from SVN and compares it with the TGZ 
files, so why does it do it for POMs? In both cases the src.tgz file is enough. 
The change to fix this should not be to hard, because the src.tgz was already 
extracted by smoker (the one of Solr is ideal, as it contains everything). 
Unfortunately my very very very limited / not existing Python knowledge does 
not allow me to fix this :-)

The other stuff like Maven artifact processing should indeed work as usual. I 
am not 100% sure why the nightly smoke in Jenkins works, but I think it may 
work because the branches are still in SVN (just older versions).

> smokeTestRelease.py's maven checker needs to switch from svn to git
> -------------------------------------------------------------------
>
>                 Key: LUCENE-7024
>                 URL: https://issues.apache.org/jira/browse/LUCENE-7024
>             Project: Lucene - Core
>          Issue Type: Bug
>            Reporter: Michael McCandless
>
> The {{checkMaven}} function in the smoke tester seems to be loading known 
> branches from SVN to locate the branch currently being released and then 
> crawling for {{pom.xml.template}} files from the svn server.  We need to 
> switch this to crawling git instead, but I'm not too familiar with what's 
> happening here ...
> Maybe [~steve_rowe] can help?



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org

Reply via email to