[OSM-talk] What is 'Attic Data'? or 'Why can't wiki writers use plain language'.

Roland Olbricht roland.olbricht at gmx.de
Fri Feb 5 06:58:47 UTC 2016

Hi all,

> Why is it headed as 'Attic Data'? Why is there no explanation of it's
> meaning? After too much time using google I find it means old
> information. Why not say that? Wouldn't 'historical snapshot' be a bit
> clearer?

It is on purpose not named 'historical'. It would have become a homonym, 
and having homonyms is always asking for trouble.

There is a whole project, named OpenHistoricalMap that collects data of 
features that have existed in human history, like Roman streets. That's 
history. It's a firm definition we cannot and should not overturn. In 
particular, OpenHistoricalMap does use for a good reason tags like 
start_date and end_date independent of when a representation is written 
to the database.

What we have here are 'outdated representations' of more or less 
unchanged objects on the ground. It may sometimes reflect changes on the 
ground, but most of changes happen due to mapping refinement or 
vandalism. Hence, we need a different notion to express that we refer to 
representations that were current at an earlier point in time.

It is then indeed reused from the well-known VCS terminology: a VCS 
keeps as well the extra data necessary to turn the controlled system 
back to consistent state at an earlier date. That kind of data is called 
"attic" there, and it is precisely the same concept as we have here.

A fun fact: the whole attic feature of Overpass is designed on the same 
decisions as the CVS storage. It turns out to be quite useful. We are as 
disk-space-constraint now as the users were for source code at the time 
CVS was developed in 1990.

Best regards,


More information about the talk mailing list