Re: [Gnumed-devel] Server vs CVS options to increase gnumed input

[David Grant]

James Busser wrote:

> The question of getting broader input (and good ideas) into gnumed 
> overlaps with making access easier.
>
> There was previous discussion of a public server at:
> http://mail.gnu.org/archive/html/gnumed-devel/2003-04/msg00019.html
>
> Would a public server allow people to participate by having to 
> download only a client? 

I think Karsten already has one.  I have accessed it before, a few 
months ago.  There was no data in it I don't think, and a few major 
things were unusable at that time.  Sadly I haven't tried the client 
only using the public server again, but I really should try again and 
give feedback.

> It is possible that, having downloaded the component software needed 
> for the client, it is felt a "simple" matter to "only" download the 
> cvs and set up one's own machine as both client and server but this 
> can be fiddly and might cost us participation.

Yes, forcing everyone to download client and server will cost gnumed 
participation for sure.  Running postgres in windows with cygwin is too 
hard for n00bs, and even for me, setting up postgres is Linux is a bit 
of a challenge, since I don't know much about databases.  (However since 
I discovered webmin it seems much easier.  I should try again.  :-) ).

> I am wondering if further a server would help people examine and 
> discuss the same thing rather than behaviour that might be based on 
> different or non-updated versions of the server.
>
> If there is some downside to this, then I wonder if there is a way 
> around some of Richard's constraints. I am here talking about CVS as I 
> try to better understand it. I am imagining that even if Richard has a 
> version of gnumed installed, Ian and Karsten's experimentations are 
> done in their own respective sections of the CVS.

Their experimentations should be done in their "working copy" of CVS on 
their own hard drive.  Once it is tested to work properly, it should be 
committed to the tree.  All the commits will merge themselves together, 
and even the working copy can be updated from other people's commits by 
running "cvs update".  I don't recommend anyone making experimentations 
in different parts of the "repository". 

> Is it possible, practical or too prone to accidental overwriting for a 
> person to maintain on their own computer not only the "current / 
> default" version of gnumed, plus their own experimental changes, but 
> to also download someone else's test code into yet another area on 
> their own computer? 

They should be able to get someone else's test code (which was deemed 
stable and committed to the repository) by doing "cvs update".  This 
will merge the other person's commits into your local working copy.  So 
you have your own "current/default" as you say + the committed changes 
of someone else.  If someone else did some real crazy stuff and it is 
not ready to commit but they want to show it to you.... they should do a 
"cvs diff -u" of the whole tree and email it.  Or email the working dir 
and put it up for download somewhere.

> That way if Iain or karsten were to say:
>     "Here, I have changed / added / removed this, what do you think"
> it might be possible to run the "ian" or "karsten" version and discuss 
> on the basis of having been able to try it out?

> _______________________________________________
> Gnumed-devel mailing list
> address@hidden
> http://mail.gnu.org/mailman/listinfo/gnumed-devel