I believe jethro is thinking about the problem actually proposed while BrowserUk is talking about a different problem. Of course, the root node is not completely clear on this point, so it could be that BrowserUk is correct and several others who have replied are the ones who are wrong.
BrowserUk is picking a (potentially) different set of columns for each row while I believe a single list of columns is desired such that this same list of columns (in the same order) provide a unique key for each row.
With a single list of columns and a useful delimiter between values, then the order no longer matters and only 2**100 potential solutions exist. Of course, you can also envision data where there is no completely effective delimiter (unless you also provide for escaping parts of values when the key is constructed). I choose to assume either an easy delimiter choice or building the key with potential escaping such as constructing the key using standard CSV format.
- tye
In reply to Re^4: Finding Keys from data (depends)
by tye
in thread Finding Keys from data
by aartist
| For: | Use: | ||
| & | & | ||
| < | < | ||
| > | > | ||
| [ | [ | ||
| ] | ] |