Re: Forrest zone auto startup
Bertrand Delacretaz wrote: > David Crossley wrote: > > >...Forrest doesn't have automatic httpd restart and > >the zones machine goes down a lot... > > Note that, altough Solaris apparently has something new and improved, > the old rc*.d stuff does work. On the cocoon zone we have created the > usual apache2 start script links: > > /etc/rc0.d/K16apache2 > /etc/rc1.d/K16apache2 > /etc/rc2.d/K16apache2 > /etc/rc3.d/S50apache2 > /etc/rcS.d/K16apache2 > > And they work fine. > > Also, if you need to startup non-root stuff automatically, I have > created a script for that on our zone, see > http://issues.apache.org/jira/browse/COCOON-1923. This is also started > in /etc/rc3.d and calls $HOME/rc/start and $HOME/rc/stop for each user > where these files exist. We use it to start Daisy, the Cocoon demos > and Continuum on our zone, all running under their respective > usernames. Thanks for your help mate. One of us that has access to both zones will get around to it. http://issues.apache.org/jira/browse/FOR-940 -David
Forrest zone auto startup (was: [RANT] The Cocoon website: move on, nothing is happening here)
On 10/20/06, David Crossley <[EMAIL PROTECTED]> wrote: ...Forrest doesn't have automatic httpd restart and the zones machine goes down a lot... Note that, altough Solaris apparently has something new and improved, the old rc*.d stuff does work. On the cocoon zone we have created the usual apache2 start script links: /etc/rc0.d/K16apache2 /etc/rc1.d/K16apache2 /etc/rc2.d/K16apache2 /etc/rc3.d/S50apache2 /etc/rcS.d/K16apache2 And they work fine. Also, if you need to startup non-root stuff automatically, I have created a script for that on our zone, see http://issues.apache.org/jira/browse/COCOON-1923. This is also started in /etc/rc3.d and calls $HOME/rc/start and $HOME/rc/stop for each user where these files exist. We use it to start Daisy, the Cocoon demos and Continuum on our zone, all running under their respective usernames. -Bertrand
Re: [RANT] The Cocoon website: move on, nothing is happening here
On 10/19/06, Ross Gardler <[EMAIL PROTECTED]> wrote: ...But is there any need for the news item to be sent to the live server immediately? I don't think so. A daily update will do just fine. For now use the Forrest generated tar, or a maven generated tar if someone sets it up... You're right, we don't have a need for immediate publishing, a few hours delay could be good enough. Let's see what happens with continuum, but in the meantime we could work as you suggest (assuming someone does write stuff, of course ;-) -Bertrand
Re: [RANT] The Cocoon website: move on, nothing is happening here
Bertrand Delacretaz wrote: > ... > (http://forrest.zones.apache.org/ seems to be down right now, cannot > check it size but I assume it's quite big). Grrr, Forrest doesn't have automatic httpd restart and the zones machine goes down a lot. So it is manual restart. I just did it and updated the notes so other committers can do it. http://forrest.apache.org/zone.html#admin -David
[jira] Closed: (COCOON-1936) SearchGenerator does not use parameterized analyzer
[ http://issues.apache.org/jira/browse/COCOON-1936?page=all ] Jörg Heinicke closed COCOON-1936. - Fix Version/s: 2.2-dev (Current SVN) 2.1.10-dev (current SVN) Resolution: Fixed Assignee: Jörg Heinicke Patch applied. Thx for your contribution. Jörg > SearchGenerator does not use parameterized analyzer > --- > > Key: COCOON-1936 > URL: http://issues.apache.org/jira/browse/COCOON-1936 > Project: Cocoon > Issue Type: Bug > Components: Blocks: Lucene >Affects Versions: 2.1.8, 2.1.9 >Reporter: Jeroen Reijn > Assigned To: Jörg Heinicke > Fix For: 2.2-dev (Current SVN), 2.1.10-dev (current SVN) > > Attachments: SearchGenerator.java.patch > > > Eventhough the documentation on the SearchGenerator says the analyzer for the > SearchGenerator can be parameterized, the search generator always uses the > default hard-coded org.apache.lucene.analysis.standard.StandardAnalyzer. > http://cocoon.apache.org/2.1/userdocs/search-generator.html > I've modified the SearchGenerator in such a way that the analyzer can be > defined from now on. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://issues.apache.org/jira/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira
Re: [RANT] The Cocoon website: move on, nothing is happening here
Ross Gardler skrev: Daniel Fagerstrom wrote: ... Second we need some (simple) way to suggest news. I think we should suggest possible news items at the dev-list by having a special headers prefix like [news]. I'd suggest just letting (self-registered) people add a news item to Daisy. Committers items will be published automatically, others will require publication by a committer - in daisy this is just a click of a link once logged in. Posting to the list is just a step too many in my view. Why not put it straight in Daisy where it can be edited and published quickly and easily. Don't forget Daisy edits are already sent to the docs list. Sounds good, we should make it as easy as possible to contribute. If someone is unsure if some material is suitable they can ask on the list. Otherwise it is just to go ahead and publish it. I hope that people will post news that are close to being too commercial ;) For such cases we certainly will need to discuss where the limit goes. Third, as the website is our official voice, we need some kind of community oversight. I think lazy consensus should be enough. If no one have protested in maybe three days, we should add the news item to the news page. Of course if someone with marketing skills would like volunteer and take a larger responsibility for creating and editing the news contents that would great. Sure, this all works fine with direct entry into Daisy rather then on the list (where everyone and their dog will chip in but only one or two will actually do anything). Daisy can be configured to only publish items that were written x days ago, thus automatically allowing for lazy consensus. +1 /Daniel
Re: [RANT] The Cocoon website: move on, nothing is happening here
I wrote a reply to Bertrands post then read the thread and discovered Reinhard is going to set up continuum. Seems like we have a volunteer and since I'm only giving ideas not implementation time you'd all be best to ignore the rest of this post ;-) Ross Bertrand Delacretaz wrote: On 10/18/06, Ross Gardler <[EMAIL PROTECTED]> wrote: ...Which publishing tool to use? I don't care. Forrest does it well (but it does need a new skin,... Is there a way to republish only a few pages quickly from Daisy to the online site? Or how hard is that to implement? A very good question - and one of the biggest problems with Forrest site generation at this time. Right now it cannot be done. It could be implemented, sure, but it is not easy, at least with my knowledge of the Cocoon CLI and crawler. The problem is that the crawler will follow all links on the page. Daisy can output a recent changed list, perhaps this could be used to limit the pages generated. A solution that requires no tools work... The Forrest zone is already building the site once a day as part of our docs validation process. Why not just grab the tar, untar, svn up (sounds like a script to me). Without going into a tools discussion, having a quick way to update just the homepage and the news (and maybe an aggregated blog feed) would help us be more active on those pages I guess. If I can add a news item in five minutes I'll do it regularly, if it takes 15 minutes it'll be much less often. Fair point. But is there any need for the news item to be sent to the live server immediately? I don't think so. A daily update will do just fine. For now use the Forrest generated tar, or a maven generated tar if someone sets it up. It will mean that someone in the community will need to log in to the Cocoon zone, execute a script, provide their SVN username and password and logout. I can't think of a faster way without working on the tools. Of course, the problem is who is the *someone*. Might simply be a case of the Forrestbot generating a subset tar.gz instead of the one pointed to by http://wiki.apache.org/cocoon/CocoonWebsiteUpdate (http://forrest.zones.apache.org/ seems to be down right now, cannot check it size but I assume it's quite big). No need to download to your local machine, do all the work in the zone.
Re: Publishing process for cocoon.apache.org
Reinhard Poetz wrote: I will setup Continuum to publish the 2.2 docs at the zone this week. Just remember, Continuum was meant to do continuous integration. Getting it to do site publishing is doable, but IMO not a good idea. If all you need is to run a maven goal using a specific profile then you'ld be much better off with cron and some log dumping to public_html. Continuum has a fair amount of quirks, just like maven has. Not to mention the fact that i regularly have to rebuild the instance because the database gets corrupted. In any case, if you need a hand setting it up let me know. Cheers, Jorg
Re: Publishing process for cocoon.apache.org
Vadim Gritsenko wrote:
Reinhard Poetz wrote:
hepabolu wrote:
As said earlier: I volunteer to run the process to publish the site
as long as I don't have to do more than start a script, provide a
login and password and get clear error messages (preferably none at
all) that tell me where to look for the trouble.
I will setup Continuum to publish the 2.2 docs at the zone this week.
...
So the question now is: Which process do we want to establish for doc
publishing?
Can you set it up so that it generates the docs into the svn checkout?
yes, that's no problem.
The full procedure could go like this:
[Initial Setup]
1. Checkout cocoon-site with --no-auth-cache
2. Setup $TOOL to (re)generate site periodically into the checkout
directory
[Recurring Manual Task]
1. ssh zones
2. cd cocoon-site
3. svn st, svn diff as necessary
4. svn ci with --username and --password
This skips manual copying and complies with infra policies.
the "--no-auth-cache" hint is great! Thanks. I will set this up ASAP.
--
Reinhard Pötz Independent Consultant, Trainer & (IT)-Coach
{Software Engineering, Open Source, Web Applications, Apache Cocoon}
web(log): http://www.poetz.cc
___
Der frühe Vogel fängt den Wurm. Hier gelangen Sie zum neuen Yahoo! Mail: http://mail.yahoo.de
Re: Publishing process for cocoon.apache.org
Reinhard Poetz wrote: hepabolu wrote: As said earlier: I volunteer to run the process to publish the site as long as I don't have to do more than start a script, provide a login and password and get clear error messages (preferably none at all) that tell me where to look for the trouble. I will setup Continuum to publish the 2.2 docs at the zone this week. ... So the question now is: Which process do we want to establish for doc publishing? Can you set it up so that it generates the docs into the svn checkout? The full procedure could go like this: [Initial Setup] 1. Checkout cocoon-site with --no-auth-cache 2. Setup $TOOL to (re)generate site periodically into the checkout directory [Recurring Manual Task] 1. ssh zones 2. cd cocoon-site 3. svn st, svn diff as necessary 4. svn ci with --username and --password This skips manual copying and complies with infra policies. Vadim
Publishing process for cocoon.apache.org
hepabolu wrote:
As said earlier: I volunteer to run the process to publish the site as
long as I don't have to do more than start a script, provide a login and
password and get clear error messages (preferably none at all) that tell
me where to look for the trouble.
I will setup Continuum to publish the 2.2 docs at the zone this week.
We need to discuss how we want to publish the real website. This could be also
done by Continuum automatically but in this we would circumvent the SVN. If we
don't want this because we don't want to break the infra policy like many other
ASF projects, the documentation publishing process would require two local steps
("mvn site-deploy" and "svn commit") and one at people.apache.org (svn update).
So the question now is: Which process do we want to establish for doc
publishing?
--
Reinhard Pötz Independent Consultant, Trainer & (IT)-Coach
{Software Engineering, Open Source, Web Applications, Apache Cocoon}
web(log): http://www.poetz.cc
___
Der frühe Vogel fängt den Wurm. Hier gelangen Sie zum neuen Yahoo! Mail: http://mail.yahoo.de
Re: [RANT] The Cocoon website: move on, nothing is happening here
On 10/19/06, hepabolu <[EMAIL PROTECTED]> wrote:
... I'm not into the full details of the update process of the site, but if
we're only updating a few pages ("home" and "news"), is it possible to
just extract those and put them directly onto cocoon.apache.org without
going through the svn process?...
Not at this time, it has to go via svn. But if it's just downloading a
small tar.gz from the forrestbot, unarchiving, checking and committing
2-3 files it'd be quick enough I guess.
-Bertrand
[jira] Updated: (COCOON-1936) SearchGenerator does not use parameterized analyzer
[ http://issues.apache.org/jira/browse/COCOON-1936?page=all ] Jeroen Reijn updated COCOON-1936: - Attachment: SearchGenerator.java.patch > SearchGenerator does not use parameterized analyzer > --- > > Key: COCOON-1936 > URL: http://issues.apache.org/jira/browse/COCOON-1936 > Project: Cocoon > Issue Type: Bug > Components: Blocks: Lucene >Affects Versions: 2.1.8, 2.1.9 >Reporter: Jeroen Reijn > Attachments: SearchGenerator.java.patch > > > Eventhough the documentation on the SearchGenerator says the analyzer for the > SearchGenerator can be parameterized, the search generator always uses the > default hard-coded org.apache.lucene.analysis.standard.StandardAnalyzer. > http://cocoon.apache.org/2.1/userdocs/search-generator.html > I've modified the SearchGenerator in such a way that the analyzer can be > defined from now on. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://issues.apache.org/jira/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] Created: (COCOON-1936) SearchGenerator does not use parameterized analyzer
SearchGenerator does not use parameterized analyzer --- Key: COCOON-1936 URL: http://issues.apache.org/jira/browse/COCOON-1936 Project: Cocoon Issue Type: Bug Components: Blocks: Lucene Affects Versions: 2.1.8, 2.1.9 Reporter: Jeroen Reijn Eventhough the documentation on the SearchGenerator says the analyzer for the SearchGenerator can be parameterized, the search generator always uses the default hard-coded org.apache.lucene.analysis.standard.StandardAnalyzer. http://cocoon.apache.org/2.1/userdocs/search-generator.html I've modified the SearchGenerator in such a way that the analyzer can be defined from now on. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://issues.apache.org/jira/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira
Re: [RANT] The Cocoon website: move on, nothing is happening here
On 10/19/06, hepabolu <[EMAIL PROTECTED]> wrote: ...Let's focus on improving/extending the actual documentation content first. Just to clarify my understanding of the current discussion: IIUC, Arje's point is to make the homepage and one or two news pages much more lively, and concentrate our communication to http://cocoon.apache.org The other docs also need to be improved as well, of course, but that's more of a medium-term goal I guess, given the current limited manpower available to do that. -Bertrand
Re: [RANT] The Cocoon website: move on, nothing is happening here
On 10/18/06, Ross Gardler <[EMAIL PROTECTED]> wrote: ...Which publishing tool to use? I don't care. Forrest does it well (but it does need a new skin,... Is there a way to republish only a few pages quickly from Daisy to the online site? Or how hard is that to implement? Without going into a tools discussion, having a quick way to update just the homepage and the news (and maybe an aggregated blog feed) would help us be more active on those pages I guess. If I can add a news item in five minutes I'll do it regularly, if it takes 15 minutes it'll be much less often. Might simply be a case of the Forrestbot generating a subset tar.gz instead of the one pointed to by http://wiki.apache.org/cocoon/CocoonWebsiteUpdate (http://forrest.zones.apache.org/ seems to be down right now, cannot check it size but I assume it's quite big). -Bertrand
Re: [RANT] The Cocoon website: move on, nothing is happening here
Ross Gardler said the following on 18/10/06 23:36: Daniel Fagerstrom wrote: [news items] Guys, I wholeheartily welcome your ideas of news items and I fully agree with Ross that they should be entered in Daisy directly. If Daisy can be configured to auto-publish after a delay, that's great. Daisy can be used to create "blog like" pages that can be automatically brought together into a news page. I agree that it should be the home page, but Daisy would not limit the info to just this page. Perhaps 3 items on the home page, and a larger news only page. Note that Daisy can also be made to create RSS feeds, but that's a "next step". I was thinking right along these lines, so I fully agree. Alternatively, have the site generation pull content from peoples existing blogs. Forrest has a plugin for this (although it is pretty basic), I'm sure Maven can be made to do it. The problem with this approach is that there is no control over the content that is published. I agree, I'd rather have a list of blogs of Cocoon committers/users than pull content directly from their blog. This, however, is of secondary importance. Let's focus on improving/extending the actual documentation content first. Of course there are lots of other ways, but they involve new tools so I'm steering away form those. Thanks for not going into tools discussions. First I think we need some common idea about what is a news item. Some suggestions would be: All your suggestions look just fine, I'm sure having an exhaustive list is impossible, but your list is a great starting point. I'm more +1 concerned about *who* will write these items and who will publish the site frequently. It really is a case of providing a login and password to a publishing tool after it is configured. As said earlier: I volunteer to run the process to publish the site as long as I don't have to do more than start a script, provide a login and password and get clear error messages (preferably none at all) that tell me where to look for the trouble. Posting to the list is just a step too many in my view. Why not put it straight in Daisy where it can be edited and published quickly and easily. Don't forget Daisy edits are already sent to the docs list. True, as is pointed out earlier posting news to the dev list takes more work because it needs to be copy and pasted into Daisy, while you can enter it in Daisy straight away and be done with it. I support the above as a "small step", I think it may encourage more people into using Daisy a little. Great points. Thanks. Bye, Helma
