Hi folks, I am running an offline Rails app that periodically connects to the Internet to download the latest catalogue, and upload any new orders. I am about to implement the synching bit - and was thinking whether there is any other solution other than writing webservices and code to do a manual sync of the data, or whether there is a more nifty and robust way using say replication? Anybody done anything like this before? Joerg
on 2006-01-12 16:59
on 2006-01-12 17:36
Well, if the structure permist why not just use ActiveRecord to handle all of that? Just change the model's database connetion. Assuming the models are the same between offline and online clients this shouldn't that difficult. Check out the ActiveRecord rdocs for an example of having a model use a different database. The wiki has a few entries concerning the topic too. You of course would have to setup the syncing routine and all that good stuff, but you would have to do that anyways :) -Nick
on 2006-01-12 18:18
Not sure that this'll work, as the client is on a separate computer in a different country ... it won't be possible for me to setup an SSH connection for them to allow their app to speak directly to the remote db ... the app is distributed on CD to various people who want it ...
on 2006-01-12 19:11
Hello, You could do that with a webservice, also written in Rails. You do not need to expose a mysql-server this way. The CD-Application connects to the webservice which intelligently transmits only the missing data to the client. The action simply selects the data-rows where updated_at > syncronized_at. regards, Helmut Am 12.01.2006 um 18:18 schrieb Joerg Diekmann:
on 2006-01-12 19:32
Yeah - that's what I started doing - was just checking if there was an easier solution ... especially synching up all the primary keys from all the different CD's creating their own orders etc .... :-( Yikes.
on 2006-01-12 20:26
On Jan 12, 2006, at 8:35 AM, Nick Stuart wrote: > good stuff, but you would have to do that anyways :) > > -Nick Be very careful if you decide to go this route. you cannot currently obfuscate ruby code so your main server's database password will be available to anyone you give the application to. Cheers- -Ezra Zygmuntowicz Yakima Herald-Republic WebMaster http://yakimaherald.com 509-577-7732 firstname.lastname@example.org
on 2006-01-14 11:22
Since Rails uses integer IDs instead of GUIDs, how are you going to keep the IDs from colliding? (I'm assuming there's more than 1 user.) Here are a couple of links that explain synchronization/replication, but they are meant to be used by sysadmins: sqlylog http://www.webyog.com/ Using MySQL's Built-In Replication To Maximize Availability http://www.phpbuilder.com/columns/tanoviceanu20000...
on 2006-01-14 12:47
GUIDs would have to be the way to go. The only drawback I see of using GUIDs is that the URL's become very ugly. To use GUIDs you'd have to make your id column a 32 character string, and instead of allowing ActiveRecord to use the DB to generate the autoincrement id's, you set the id GUID value yourself in your model before saving. I haven't yet implemented it - just deducing that these could be the steps to make Rails use GUIDs instead of int. You could use http://rubyforge.org/projects/uuidtools/ to generate your GUIDs. Any thoughts?
on 2006-01-14 14:38
Yes, nasty ugly URLs. I'm afraid I don't have any experience with GUIDs since IDs have been sufficient for my needs. I was just interested in whether you had some new, super-cool way of merging. You might also check out, http://dema.ruby.com.br/articles/2005/09/14/guid-u... and Typo also uses GUIDs in a few places. Lots of luck.