[OPEN-ILS-GENERAL] generic training server with predictable data

[Dan Scott]

On Mon, 2010-06-21 at 17:09 -0700, Lori Bowen Ayre wrote:
> Hi All,
> 
> I posted info about this idea and got a big 'ol nothing in response so
> I wanted to try one more time to verify that it really is of no
> interest to anyone. 

It is of interest, I just have a slight variation on your suggestion
which really hearkens back to a suggestion I pitched many moons ago. (I
think the first time might have been
http://markmail.org/message/dbwre7cqhmmnnlrv but I know I've discussed
it in several other forums since then...)

>  Here's the idea....as one of the RSCEL projects, were thinking of
> setting up a training server which would have a current version of
> Evergreen loaded, would be accessible to anyone (much like the demo
> servers) but we'd make sure we kept it totally current with one
> matching client software download.  
> 
> 
> We'd also keep data in there that you could rely on being there for
> training (so we'd have to refresh it on some regular basis).
>  Eventually, we'd contribute some exercises that people could use in
> their training which would use that training server and data.  
> 
> 
> The idea is that you would then not have to worry about how to train
> your people on your own Evergreen server at the same time you were
> trying to get ready to go live.  
> 
> 
> Also, by sharing the same data and server set-up, we could all
> contribute training exercises.
> 
> 
> So, if you think this would be useful or you think it would be a waste
> of time...please advise.  Friendly amendments are also encouraged.  We
> want to do something that helps people so if this wouldn't help
> anyone, we RSCELs will focus our attention somewhere else!
> 

Rather than having this data sitting just on a training server RSCEL, I
think it would be much more useful to have the sample data sets
available as part of the Evergreen source. This would enable every
Evergreen install to (optionally) have a reference set of data to work
with locally. We could build documentation / training based on those
sample sets of data, but also build tests on those sample sets of data
to ensure that the SQL schema hasn't been broken & the data can actually
load, a given server is working as expected, and that a given report
returns the expected results, and that a given API call generates the
expected effects.

This could also be a useful basis on which to build tests of the
migration scripts.

I do think that having an up-to-date training server always at the ready
would be a valuable thing and support that goal; I just think that it's
potentially more important to have a canonical set of sample data
available as part of the Evergreen source.