[Geomoose-users] OSGeo Incubation progress

[Jim Klassen]

On May 10, 2012, at 5:36, Jody Garnett <jody.garnett at gmail.com> wrote:

> This discussion is moving along at a great rate; could we capture the "answers" in a wiki page as they are sorted out? I am really keen to get through this promptly; and we can ask for any feedback in Monday's incubation meeting.
>>>  In any case, our docs are "open".  I think OSGeo's main concern is preventing things like (past versions of) numpy that had free code, but no docs, unless you paid.
>> So at least you and I think that we should license our docs under our existing code license. In all cases, they are certainly 'open'.
> If you have this sorted; I am interested in writing down what is in place today. We can always hunt down someone with more experience if we need guidance on where to go tomorrow.
Ideas? My take would be MIT for everything in SVN (incl website).  Wiki is under what terms?

>>> Is the #geomoose IRC room archived somewhere?
>> No. I have some logs locally on various computers but have made no
>> attempt to really log it. If the channel is active, we could ask Gary
>> Sherman to send his bot and post the archives as he kindly does for
>> some other OSGeo Projects.
> We post the geotools logs to a blog or something whenever we have a formal IRC meeting.
> Indeed the most popular way to end a meeting was to offer to post the logs.

Ah, we could post IRC logs next to our (or as part of our) published meetings minutes. 

> 
>>>>>   The project follows a documented testing process.
>>>>>   Ideally, this includes both automated and manual testing
>>>> 
>>>> More than Incubation, this could really help us.  I have some mediocre
>>>> ideas on this (like bash/curl/diff/etc to make autotests), if others
>>>> have good ideas, I'd be interested in pursuing that.
>>> 
>>> The "hard" part is what I'd really like to automatically test is the JavaScript part across multiple browsers.  Is there such a thing as a build-bot for a browser based app?
> Consider focusing on what you do for testing today?
> 
> Minimal:
> - Make an experimental release (often called release candidate) and ask the community to test for a week. if they don't complain loud enough publish it as the release.
> 
> Manual:
> - a couple of sample data sets are put through their paces manually. Often this is a good approach if you have training material, or installation guides to start from.
> 
> Maximal:
> - conformance tests provided by a third part (example OGC "Cite" tests available for several of their web services; and possibly some clients)

We do a combination of the minimal and manual testing right now.