Hi Rob, take a look at the 3rd clause of
couch_db_update_notifier:handle_event (line ~55). That's the code
that sends update notifications to your Ruby script. It handles
couch_db_update_notifier:notify() calls, which you'll find in
couch_server.erl and couch_db_updater.erl.
From a technical perspective it would be easy enough to change the
"updated" atoms sent in couch_db_updater to something a little more
detailed (and extend the gen_event code to work with terms other than
atoms at the same time). I haven't used DbUpdateNotificationProcesses
much myself so I'm not sure if there are higher-level design reasons
why sending this more detailed info should be discouraged. Best,
Adam
On Dec 4, 2008, at 4:51 PM, Rob Law wrote:
Hey,
I am trying to modify the data that gets sent to my ruby script when
an
Update happens (as described here :
http://wiki.apache.org/couchdb/Regenerating_views_on_update). I am
hoping
to have couchdb alert me with the actual document that was updated
or added
(like id, and attributes, etc), so that I can parse this update in
my ruby
script and act accordingly. So instead of the update_call looking
like:
{"type":"updated","db":"DB_NAME"}
it would look like this:
{"type":"updated","db":"DB_NAME", "document":"_id" : "2342331232",
"data":"test"}
Is there any hope of getting that kind of data from couch on document
update? I could just have my ruby script request changed documents or
something, but I thought it would be cleaner and more extensible if
couch
just told my script the document that got updated.
Any tips would help. Also, could someone direct me to the file and/
or line
where that data gets sent to my script. This is my first foray into
Erlang
and I am trying to get into playing with couches internals, etc.
thanks
Rob
