... the world's cruftiest CGI.
http://www.cantrell.org.uk/david/tech/openpostcode/
I've started off with a few places, others can add as they see fit. If anyone wants to take the code, prettify it, fix bugs, write documentation, make it compliant with the latest fashion in HTML and templating, host it somewhere, or knit it into a nice tea cosy, please do.
On Sun, Oct 26, 2003 at 01:02:54PM +0000, David Cantrell wrote:
... the world's cruftiest CGI.
Hmm. Anyone who knows more about database right know whether recreating this data is database right theft (the same question can potentially be asked of using streetmap or similar to get the lat/longs for entry into Openguides directly)
Cheers,
Dominic.
On Sun 26 Oct 2003, David Cantrell david@cantrell.org.uk wrote:
What are you planning to do with the data? How are you going to prove that you didn't nick it from streetmap?
Kake
On Mon, Oct 27, 2003 at 10:43:34AM +0000, Kate L Pugh wrote:
On Sun 26 Oct 2003, David Cantrell david@cantrell.org.uk wrote:
What are you planning to do with the data?
Right now? Not a lot. With enough data, it would enable people to do things like list all the FOOs within N distance of place M, given the addresses of M and the FOOs.
How are you going to provethat you didn't nick it from streetmap?
Streetmap are as welcome to look at the source code as you are. I've checked the code for Geography::NationalGrid too, to make sure that's not doing anything naughty.
The real reason I did this is because people talked about it and there was some vague muttering about registering a domain, but no-one DID anything. I thought "ooh, that might be useful", so did it.
On Mon, Oct 27, 2003 at 01:23:24PM +0000, David Cantrell wrote:
Streetmap are as welcome to look at the source code as you are. I've checked the code for Geography::NationalGrid too, to make sure that's not doing anything naughty.
The thing is, sure they can comb through the code, but there's nothing that proves the source of the data itself. However, not being a lawyer, I don't know on whom the burden of proof would be if we were taken to court - the plaintiff, or the defendants? I don't see how either side could prove anything. Anyone here know a lawyer?
The real reason I did this is because people talked about it and there was some vague muttering about registering a domain, but no-one DID anything. I thought "ooh, that might be useful", so did it.
It is useful; thank you.
Earle Martin wrote:
On Mon, Oct 27, 2003 at 01:23:24PM +0000, David Cantrell wrote:
Streetmap are as welcome to look at the source code as you are. I've checked the code for Geography::NationalGrid too, to make sure that's not doing anything naughty.
The thing is, sure they can comb through the code, but there's nothing that proves the source of the data itself. However, not being a lawyer, I don't know on whom the burden of proof would be if we were taken to court - the plaintiff, or the defendants? I don't see how either side could prove anything. Anyone here know a lawyer?
Yes, I realised on the way home this evening that kake had probably meant what proof was there that users hadn't just cut n' pasted data from streetmap. And of course there's no way for me to be able to tell.
Of course, the whole of Openguides suffers from this. Who can tell whether some random review is cut n' pasted from (eg) the Evening Standard's website?
And I think I owe an apology. I was far too grumpy when I posted this afternoon.
[ http://www.cantrell.org.uk/david/tech/openpostcode/ ]
On Mon, Oct 27, 2003 at 10:43:34AM +0000, Kate L Pugh wrote:
What are you planning to do with the data?
On Mon 27 Oct 2003, David Cantrell david@cantrell.org.uk wrote:
Right now? Not a lot. With enough data, it would enable people to do things like list all the FOOs within N distance of place M, given the addresses of M and the FOOs.
OK, but the problem with the data attribution (see below) would need to be solved before I'd consider relying on this for any application I was involved with writing.
Kake:
How are you going to prove that you didn't nick it from streetmap?
Dave:
Streetmap are as welcome to look at the source code as you are.
But that doesn't prove anything about where the data came from. If Streetmap - or anyone else who licenses postcode data - can make a convincing case that some of the data in your collection was harvested from their site, then you run the risk of not being allowed to distribute *any* of your carefully-collected data. And then any applications which depend on your data will break.
The real reason I did this is because people talked about it [...] but no-one DID anything.
People who have thought about this kind of thing fairly carefully include blech, jerakeen and Chris Heathcote. I'd recommend you have a chat to at least one of those people if you plan to put much effort into this project.
The direction I am going in is the one you and others suggested on london-pm - the simple consideration that things in SW6 are going to be fairly close to things in SW6 6LA. If CGI::Wiki's tests hadn't been designed by someone with bananas for brains (ie me) then this would be done by now.
Kake PS - I'm sure you know this, but postcodes do change.
[http://www.cantrell.org.uk/david/tech/openpostcode]
On Tue, Oct 28, 2003 at 12:44:09PM +0000, Kate L Pugh wrote:
On Mon 27 Oct 2003, David Cantrell david@cantrell.org.uk wrote:
On Mon, Oct 27, 2003 at 10:43:34AM +0000, Kate L Pugh wrote:
What are you planning to do with the data?
Right now? Not a lot. With enough data, it would enable people to do things like list all the FOOs within N distance of place M, given the addresses of M and the FOOs.
OK, but the problem with the data attribution (see below) would need to be solved before I'd consider relying on this for any application I was involved with writing.
See also Openguides and its content, where you have no way of knowing whether it's cut n' pasted from $payware_restaurant_review_site or copied from a book.
The real reason I did this is because people talked about it [...] but no-one DID anything.
People who have thought about this kind of thing fairly carefully include blech, jerakeen and Chris Heathcote. I'd recommend you have a chat to at least one of those people ...
[CCed]
I don't see that it requires much thought really. It's a very simple database. There's no *sensible* reason for it to be considered naughty.
PS - I'm sure you know this, but postcodes do change.
Yup. Thankfully they don't change often, or by much, and in comparison to the number of codes, the number of changes is infinitesimal. ie, for the purposes of figuring out where your nearest pub is, we don't have to care.
On Tue, Oct 28, 2003 at 02:18:08PM +0000, David Cantrell wrote:
See also Openguides and its content, where you have no way of knowing whether it's cut n' pasted from $payware_restaurant_review_site or copied from a book.
Only slightly; it's easy to prove something like a review was copied - compare it against the original. Reviews have a high uniqueness level[0], and running a diff against a review and a possible copy will quickly expose if it's original or not. Postcodes, on the other hand, are a very short alphanumeric string for which provenance is almost(?) impossible to determine.
[0] I think the cryptographers call this "entropy".
Earle Martin wrote:
On Tue, Oct 28, 2003 at 02:18:08PM +0000, David Cantrell wrote:
See also Openguides and its content, where you have no way of knowing whether it's cut n' pasted from $payware_restaurant_review_site or copied from a book.
Only slightly; it's easy to prove something like a review was copied - compare it against the original. Reviews have a high uniqueness level[0], and running a diff against a review and a possible copy will quickly expose if it's original or not. Postcodes, on the other hand, are a very short alphanumeric string for which provenance is almost(?) impossible to determine.
Mmmm, fair point.
However, they wouldn't be bitching about the postcodes, but about the map references related to them. I just checked two of my data points. In both cases both the easting and northing are different to what streetmap has. Understandable, because postal codes cover either one very large place (such as a campus) or several smaller places (such as a row of houses), and of course because it's a matter of judgement which hundred meter division is closest to a chosen point.
openguides-dev@lists.openguides.org
-
David Cantrell -
Dominic Hargreaves -
Earle Martin -
Kate L Pugh