[OSGeo-Discuss] Open Location Services

[Jody Garnett]

> The above makes sense, but honestly, I had never heard of this until
> now, and I have been tinkering with open source for almost a decade
> now. Most open source projects seemed organic to me. Someone had an
> itch, they scratched it, they put it out, and the project either
> gathered traction, or it died. Seems like my scholarship of open
> source has been lacking in this aspect hugely.

You will notice that I expressed the same concern from the other side
of the street. When GeoTools is approached by a private company with
functionality to "donate" we are often forced to say no.

A couple of examples:
- the jts-wrapper module was developed in 2004 offering ISO 19107 api
wraped around the jts library. We accepted the library as an
unsupported module in 2005, taking the time to add it to our build etc
 - and nothing has happened with it. As a result we will shortly be
removing it.
- OpenLS was offered but without community resources; or any requests
from our user list asking for the functionality; we were forced to say
no
- the geometry module is a full fork of jts backed on to ISO 19107
api. Developed by a grad student and donated to the library; this time
the university was on hand to help adding it to the build; and
commercial funding was received to document the result. Still without
an active maintainer uptake has been low and we will be removing it.

> I wonder if I can find out the "packaging costs" of other projects,
> for example, what was the packaging cost for MapServer, or GeoServer,
> or OpenLayers, or Perl/Python, etc.

These situations are often slightly different in that open source was
used as venue to enable collaboration between several interested
parties. In the case of OpenLS we don't have any interested parties.

> Is this routine practice, or is this a consideration only when a
> private company wants to put its code into open source?

The cost of making code open source is measured in volunteer time;
either someone such as yourself who is interested in the
functionality; or an organisation paying their employees to do the
work.

Phrased another way a private company can benefit from making its code
open source if it can attract a larger user community to cut down on
testing costs.

I recently "ran" these numbers for GeoServer. It costs LISAsoft a
couple of days time to release the most recent GeoServer (thanks to
LISAsoft for allowing us to take part!).

In trade we got back around 4-5 useful bug reports. I figure each bug
report represents someone downloading; installing; configuring (1-2
hours) and then taking the time to write a bug report; for a useful
bug report that will take 1 hour).

So 4 reports; 3 hours each = 12 hours. It took 16 hours to make the
release - so from the standpoint of time saving it would of been
cheaper for us to just do the testing ourselves.

Now I am sure many more people downloaded and tried out the
application; but if they did not report a bug the community is not
benefiting from the release. We depend on this in order to maintain
open-source's reputation for quality. The number of bug reports
against GeoServer 2.0 vs the RC issued the week before really
highlights the need to talk about this relationship between open
source projects and the value of testing.

> Another question -- if you don't put the code into open source, are
> you somehow recouping this cost? In other words, does putting the code
> into open source have any opportunity costs? Asked another way, if you
> did just "dump the code into sourceforge," besides the potentially
> legitimate worry that the project might just die, would you incur any
> other cost?

I don't think as professionals we could just dump the code on source
forge; it would harm our reputation.

> No, I don't have any sponsor. I am a rather indigent
> academic/developer/activist with barely funds to keep myself afloat. I
> am, however, still very curious about the magnitude of these
> "packaging" costs. What are we talking about here? A few hundred, a
> few thousand, a few tens of thousands, say, Euros (considering even
> Kanye West doesn't want greenbacks anymore). You say above, "we can
> look into it further." Does that imply that you haven't yet calculated
> these packaging costs, but have a sense that they might be
> substantial?

It may also be measured in terms of man hours; we could make a copy of
the code available to your organisation under an open source license
(but that would not be the same as an open source project that you
could take part in).

This is some times seen as the difference between "open source" and
"open development". OSGeo is focused on an open development model; we
take pains to ask projects to document their decision making process
and how new organisations and members can take part.

> At the very least, because of this thread, I have now been made aware
> of a potential aspect of open source about which I had absolutely no
> idea until now.

Don't get hung up on the phrase "packaging costs"; instead focus on
the larger picture of the community that forms around a project and
what it takes to make that relationship healthy (each party needs to
get a benefit out of it that is more then they are putting in etc...).

Regards,
Jody