| 1 |
Some questions: Computing intersections and inclusions between sets |
| 2 |
and ranges seems tricky. For instance, do one need to handle |
| 3 |
|
| 4 |
(* set (* range numeric (ge #00#) (le #20#))) |
| 5 |
|
| 6 |
includes |
| 7 |
|
| 8 |
(* range numeric (ge #05#) (le #15#)) |
| 9 |
|
| 10 |
What if certificate reduction ends up with more than one reduced 5 |
| 11 |
tuple? I think the function spki_process_sequence_no_signatures (and |
| 12 |
spki_process_sequence, when that is implemented) should return a list |
| 13 |
of 5-tuples, *and* a principal that is the ultimate subject of the |
| 14 |
sequence. I.e. either the final expression of the sequence, if that is |
| 15 |
a public-key or a hash, or the subject of the final certificate in the |
| 16 |
sequence. |
| 17 |
|
| 18 |
Next, the spki_5_tuple_reduce function should return a set of acl:s, |
| 19 |
that all have the principal in question as subject. It could also be |
| 20 |
restricted further, by giving it a date and/or a tag as argument. |
| 21 |
|
| 22 |
All this implies that we need reference counts and cons-lists to handle |
| 23 |
acl lists, as a single acl may be on more than one list. |
| 24 |
|
| 25 |
|
| 26 |
Implement range, and figure out if and how to support all the |
| 27 |
different range types. |