[postgis-devel] Re: [postgis-users] how to identify the_geom that cause : RelateOperation called withLWGEOMCOLLECTION type

[Mark Cave-Ayland]

Kevin Neufeld wrote:

> Yup, that's makes sense ... it's similar to what I have in 
> errorIfGeometryCollection.
> 
> Doing a little testing, however, I'm not clear on what you were 
> expecting to see based on certain inputs - my expected results don't 
> match the function output.  I put the function in a little test.c file 
> to test the method where:
> 
> 
> char* str = "abcdefghijklmnopqrstuvwxyz";
> 
> // Assuming 0-based indexing
> lwmessage_truncate(str, 0, strlen(str)-1, 26, 0);
> expected: "abcdefghijklmnopqrstuvwxyz"
> output:   "abcdefghijklmnopqrstuvwxy"
> 
> lwmessage_truncate(str, 0, strlen(str)-1, 26, 1);
> expected: "abcdefghijklmnopqrstuvwxyz"
> output:   "abcdefghijklmnopqrstuvwxy"
> 
> lwmessage_truncate(str, 0, strlen(str)-1, 25, 0);
> expected: "...efghijklmnopqrstuvwxyz"
> output:   "abcdefghijklmnopqrstuvwxy"
> 
> lwmessage_truncate(str, 0, strlen(str)-1, 25, 1);
> expected: "abcdefghijklmnopqrstuv..."
> output:   "abcdefghijklmnopqrstuvwxy"

These we caused by an off-by-one error in the code, as in the original 
commit endpos was actually 1-based indexed, rather than 0-based like 
startpos :(

> lwmessage_truncate(str, 2, strlen(str)-1, 10, 1);
> expected: "cdefghi..."
> output:   "abcdefg..."
> 
> lwmessage_truncate(str, 4, strlen(str)-1, 10, 1);
> expected: "efghijk..."
> output:   "abcdefg..."

Ahhh. Forgot to add the startpos offset when performing start truncation...

> lwmessage_truncate(str, 0, strlen(str)-1, 10, 0);
> expected: "...tuvwxyz"
> output:   "...stuvwxy"
> 
> lwmessage_truncate(str, 0, strlen(str)-3, 10, 0);
> expected: "...rstuvwx"
> output:   "...qrstuvw"
> 
> lwmessage_truncate(str, 0, strlen(str)-5, 10, 0);
> expected: "...pqrstuv"
> output:   "...opqrstu"

These are again a symptom of the off-by-one error.

> lwmessage_truncate(str, 0, strlen(str)-1, 4, 1);
> expected: "a..."
> output:   "a..."
> 
> lwmessage_truncate(str, 0, strlen(str)-1, 3, 1);
> expected: "a.."
> output:   "..."
> 
> lwmessage_truncate(str, 0, strlen(str)-1, 2, 1);
> expected: "a."
> output:   "abcdefghijklmnopqrstuvwxyz..."
> *** glibc detected *** free(): invalid next size (fast): 0x08b88028 ***
> Aborted

These really lie in the category of "don't do that", but they shouldn't 
invoke a crash. For the moment, anything that is truncated where 
maxlength is < 3 will simply return a fixed "..." string.

> Am I using it correctly?

Yes, of course. Fixes for all of the above have been committed to SVN - 
thanks for testing :)


ATB,

Mark.

-- 
Mark Cave-Ayland
Sirius Corporation - The Open Source Experts
http://www.siriusit.co.uk
T: +44 870 608 0063