(please ignore previous post, sending a fixed version)
oh, yes, sorry, I described option 2.
Then option 1 will require infra to work on integrating your templating engine as external task : someone knowing this engine, its prerequisites and so on has to work with infra to help them run the tool by hand then be able to launch it from build system (with source files in content directory, output directory set by a command line attribute, and and everything in a trunk directory)
Notice that if you explain infra that /whatever/file.html is generated from /content/pages/whatever/file.twig, it should be easy for infra to add support for online content editing = a really nice feature of the CMS. You can even add a "CMS boormarklet" in your bookmarks, and jump into content editing in one click from browse. This one is very nice. If this is not clear, I can explain more this feature, because the "CMS bookmarklet" isn't something everybody knows.
Le lundi 21 mai 2012 23:11:47 Ralph Goers a écrit :
Wait - I thought option 1 used the CMS and could invoke the site build.
On May 21, 2012, at 1:55 PM, Hervé BOUTEMY wrote:
seems a good step, IMHO, because your main site has its own templating engine that won't be used by many people, then don't deserve CMS dedicated code to integrate it: pure svnpubsub will be the simplest choice for everybody
and as done in the main site source, a pom.xml can provide html publish to svn: just need to check if excludes works well to avoid deleting components subsites when publishing main site update
Le dimanche 20 mai 2012 15:15:46 Ralph Goers a écrit :
Based on your other answers I'm in favor of going with option 1. I think we should start by having infra set up a loggingtest site. and then see if we can't get Ivan's build to work (preferably in English).
On May 20, 2012, at 1:28 PM, Hervé BOUTEMY wrote:
ok, so it's a custom rendering engine
so I see 2 solutions:
1. either infra adds this engine as external, like it did with "mvn site": you'll have to put sources in content, to trigger html generation on each source update, buildbot will build the html, then you'll use the CMS web interface to publish staged content
2. either infra simply adds svnpubsub, without any CMS integration, ie any source modification integration nor html build from sources. I don't know if they do that. But that way, you're completely free, you only use svnpubsub: it's up to you to get the tooling to put html content to svn.
With 1st solution, main site is automatically published at each commit, each component being protected from erase by extpath.txt.
With 2nd solution, main site is manually published when somebody does it, like components during releases, and you'll have to take care of not removing components content when publishing.
Le dimanche 20 mai 2012 12:53:27 Ralph Goers a écrit :
Here was what Ivan proposed - http://mail-archives.apache.org/mod_mbox/logging-general/201205.mbox/ajax/%3CCAKpWnhTiVEe0M54UOFHGXTUOPq3TX2Jd_Z-a7B9H_pfnxQmCDg%40mail.gmail.com%3E
On May 20, 2012, at 12:06 PM, Hervé BOUTEMY wrote:
I'm now subscribed to general@logging
Thanks Ralph for INFRA-4669: it gives me good information on actual status.
The main question IMHO for the moment is: how are you planning to generate main site html? Maven, CMS's markdown, another tool?
Then each component will have its own generation tool, with the only expectation is to output html to svn
Le dimanche 20 mai 2012 11:58:16 Ralph Goers a écrit :
If you meant me, of course I'm subscribed to Maven Dev as I'm on the PMC. However, I haven't done any work there in a very long time. This list would seem to be more appropriate for a logging related discussion.
To reiterate a bit for Hervé's sake, I've opened https://issues.apache.org/jira/browse/INFRA-4699 which is sort of in a state of limbo waiting for us to tell infra what we actually want. We haven't responded because we aren't really sure. So the first piece we need is something to tell infra so that we can actually start doing something.