Re: [Gnumed-devel] Data packs

[Karsten Hilbert]

On Thu, Oct 06, 2011 at 10:58:17AM -0700, Jim Busser wrote:

> … people working from git can (and maybe need to be) running database version
> 
>       (latest-release n) + 1

They should, at any rate, in order to expect meaningful
results. Or rather, they should use whatever the latest git
bootstraps to.

> which even the git client will show at login as n+1, as will the git client 
> menu
> 
>       Help > About database

Well, we try to deviate as little as possible from actual
code-to-be-released.

Also, what is shown there is what GNUmed *detects*, not what
it *expects*. It will then go ahead and ignore mismatches
between the two if in --debug (= devel) mode.

> Therefore it is seeming to me that setting a minimum version to v16 should 
> not prevent testing unless …

Doing so will not show the v16 minimum data packs when
running from git tree because git head *expects* a fake
database version "0".

> is the problem that v16 would not have been frozen until
> it would be officially released, and therefore any hash
> would be dynamic and (although scriptable) might be a pain
> to continually manage?

Exactly.

> In which case we could set the minimum to v16 as part of releasing v16 
> officially?

We would, indeed, as you may have noticed has been prepared in

        http://www.gnumed.de/downloads/data/data-packs.conf

since The Beginnings Of The Known Universe ;-)

Karsten
-- 
GPG key ID E4071346 @ gpg-keyserver.de
E167 67FD A291 2BEA 73BD  4537 78B9 A9F9 E407 1346