[
https://issues.apache.org/jira/browse/LUCENE-9020?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17024469#comment-17024469
]
Uwe Schindler commented on LUCENE-9020:
---------------------------------------
I was able to set everything up. It was a bit more coplicated, as AliasMatch
and absolute path names are not allowed in .htaccess files (which is a per
directory config, so actual directory is resolved). See INFRA-19439 for more
details.
We solved the issue mostly over the Slack Channel. In short what we did with
Daniel:
"Alias" and "AliasMatch" does not work in ".htaccess" (which is a per-directory
config and therefore the file system patch is already found out, so it's to
late to apply aliases. Aliases only work on server config or location). The
workaround is to use "mod-rewrite". The fix consists of 2 separate parts:
- INFRA added an Alias/Rewrite on their side in the global server config that
can be used by all project server. It maps URI path "/__root" to the filesystem
path where all project webpages are hosted: See this initial commit:
https://github.com/apache/infrastructure-p6/compare/a63511b7499f...63e23e52b18b
- Lucene/Solr changed their .htaccess to use rewrite directives that just
rewrite the above URLs to "/__root/....old-svn-website.../". This makes it
independent from real filesystem paths. We (Lucene) just know that below the
URI path "/__root" we can reach all project folders that are checked out on wb
server. Only backside: You can theoretically reach every website by
hand-crafting an URL like
https://lucene.apache.org/__root/someotherproject/somehtml. With nginx as
webserver you could define this URI path as "internal", but Apache HTTPD does
not have this notion. Nginx uses this "internal" notion for resource endpoints
only accessible by rewrites. Our htaccess now looks like this:
https://github.com/apache/lucene-site/blob/3fa9933b276897f89525c61301d9e4e2da863b85/content/.htaccess#L121-L125
We did some tests with checking out part of the SVN tree on the staging
machine. But the whole rewrite generally only works in production (which is not
different to our old website, as the old CMS was also not showing our javadocs).
The final step is to bring the website to production. We may need some more
help once this will be done, as we cannot guarantee that all works perfect in
production (we hope so).
> Find a way to publish Solr RefGuide and Javadocs without checking into git
> --------------------------------------------------------------------------
>
> Key: LUCENE-9020
> URL: https://issues.apache.org/jira/browse/LUCENE-9020
> Project: Lucene - Core
> Issue Type: Sub-task
> Reporter: Jan Høydahl
> Assignee: Uwe Schindler
> Priority: Major
>
> Currently we check in all versions of RefGuide (hundreds of small html files)
> into svn to publish as part of the site. With new site we should find a
> smoother way to do this.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org
For additional commands, e-mail: issues-h...@lucene.apache.org
