On 27 May 2017 at 03:38, Sam Ruby <ru...@intertwingly.net> wrote: > On Fri, May 26, 2017 at 10:25 PM, Shane Curcuru <a...@shanecurcuru.org> wrote: >> Optimization exercise: how would we turn the output of tools/wwwdocs.rb >> get_annotated_scan(...) into a nightly generated >> /public/whimsy-tools.json? Side effect: learning more about whimsy >> architecture. >> >> * Inputs (all public files): >> - /www/**/*.cgi where line =~ ... PAGETITLE = ... >> - >> whimsy-vm4.apache.org.yaml['vhosts_whimsy::vhosts::vhosts']['whimsy-vm-443']['authldap'] >> >> * Output (which would be sucked in by /www/committers/tools.cgi): >> - /www/public/whimsy-tools.json > > Answer to the question asked: you want a cron job. Cron jobs are > managed by puppet. The file that controls this is here: > > https://github.com/apache/infrastructure-puppet/blob/deployment/modules/whimsy_server/manifests/cronjobs.pp > > Be sure to read the infra workflow information here before proceeding: > > https://github.com/apache/whimsy/blob/master/DEPLOYMENT.md#production-configuration > > --- > > While the above is the answer to the question asked, a better answer > would be to add a call to run wwwdocs.rb to the :update task in the > Rakefile: > > https://github.com/apache/whimsy/blob/master/Rakefile#L8 > > The update task is run any time new code is deployed. > > --- > >> * Questions: >> >> - Why are the scripts that generate /public inside /www/roster, instead >> of being in /tools? Mere history, or a specific reason? > > The initial set of tools were roster related. If you want to move > them, be sure to update the relevant cronjobs. > >> - Does one *need* an ASF::module to wrap the data model, or can the >> existing generation code in wwwdocs.rb just be used as-is, wrapped in >> sufficient calls to public_json_common.rb to handle the 'write file only >> if json data changed' bits? > > Either cronjobs or rake tasks can invoke the tool as is. > >> - Better suggestions to do this? I know writing the .json to /public >> maybe premature optimization, but I'm also thinking there might be other >> uses of this data as well, meaning having a URL with the data is useful. > > Cool. It is also useful for debugging (much like the scavenge/analyze > split in the site tool).
Also HTTP URLs are easy to redirect. SVN/Git not so easy. >> -- >> >> - Shane >> https://www.apache.org/foundation/marks/resources > > - Sam Ruby
