| Type | Wish | Status | dismissed | Date | 20-Jan-2010 08:16 |
|---|---|---|---|---|---|
| Version | alpha 96 | Category | Datatype | Submitted by | PeterWood |
| Platform | All | Severity | major | Priority | normal |
| Summary | Case sensitive key option for Map! |
|---|---|
| Description |
The DocBase entry for the Map! database states that Map! keys are case insensitive. It also includes the comment "should there be a mode to allow that". Allowing a case sensitive option will increase the usefulness of Map!. It would allow building of case-sensitive word list for indexing and spell checking. It would also allow Map! to be used in building tables from the source code of case sensitive languages such as Java and JavaScript and of course, Unix file names. ANSWER: use binary! keys. They are case sensitive and the examples above would be sourced as binary anyway (as the input from files or directories). |
| Example code |
Current R3 behaviour
>> make map! ["Peter" #[true] "peter" #[true]]
== make map! [
"Peter" true
]
Desired R3 behaviour
>> make/sensitive map! ["Peter" #[true] "peter" #[true]]
== make map! [
"Peter" true
"peter" true
]
|
| Assigned to | n/a | Fixed in | - | Last Update | 4-Feb-2010 18:52 |
|---|
| Comments | |
|---|---|
|
(0001908)
pekr 20-Jan-2010 12:29 |
I am not sure there will be strong desire to overload 'make with /sensitive refinement. This one would be used only for map! dtype. Some more general aproach is needed. E.g. objects are going to have some init spec block implemented, and I suggest the same aproach to be taken for another datatypes, if there is some need .... |
|
(0001909)
abolka 20-Jan-2010 15:48 |
See also #1315. |
|
(0001910)
PeterWood 20-Jan-2010 23:45 |
I'm not at all bothered by the mechanism by which a case sensitive Map! is created - the /sensitive refinement was just a quick way to write an example. |
|
(0001911)
PeterWood 20-Jan-2010 23:52 |
Without this extension, I don't believe it is correct to state that Map! is a replacement for Hash!
Strings stored in a Hash! are case sensitive: >> h: make hash! ["peter" "Peter"] == make hash! ["peter" "Peter"] >> find h "peter" == make hash! ["peter" "Peter"] >> find h "Peter" |
|
(0001914)
BrianH 24-Jan-2010 04:13 |
The map! type is only a replacement for hash!, not compatible with it. They have different semantic models. Due to the nature of the map! data model it would be impossible to use case-insensitive addressing with case-sensitive map!, and vice versa. This means that FIND or SELECT /case doesn't work on the current case-insensitive map!, nor could it.
This ticket is a restatement of #1315 as a option. In theory it could be done with a flag similar to the not flag of bitsets. But since words can already be used in map! specs there isn't an obvious syntax method for specifying the flag the way there is with bitset! (suggestions are welcome). A /sensitive option for MAKE will not happen. MAKE options go in the spec block. MAKE is a very low-level action that won't ever get a type-specific refinement. We have to figure out another way. |
|
(0001959)
Carl 4-Feb-2010 18:50 |
It's a bit of a mind bender, isn't it?
Although it would be quite possible to allow a MAKE/options refinement to specify variations for any datatype, it becomes very problematic because we like *strong reflection* for MOLD and LOAD. This can only happen for "in-band" specifications... those that are expressed directly in the block argument passed to MAKE map!. Any /options would be lost (unless perhaps a MOLD/all format were used.) But, there is a solution for the cases mentioned above. Use binary! keys. This is less of a problem than it seems... because the input streams (a source file or file directory) are either ASCII or UTF-8, so in their raw form they're already binary! and upper and lower case will be distinguished in the key. |
| Date | User | Field | Action | Change |
|---|---|---|---|---|
| 4-Feb-2010 18:52 | carl | Description | Modified | - |
| 4-Feb-2010 18:52 | carl | Status | Modified | problem => dismissed |
| 4-Feb-2010 18:50 | carl | Comment : 0001959 | Added | - |
| 30-Jan-2010 23:30 | BrianH | Comment : 0001914 | Modified | - |
| 30-Jan-2010 02:53 | BrianH | Comment : 0001914 | Modified | - |
| 30-Jan-2010 02:51 | BrianH | Code | Modified | - |
| 30-Jan-2010 02:51 | BrianH | Status | Modified | submitted => problem |
| 30-Jan-2010 02:49 | BrianH | Comment : 0001914 | Modified | - |
| 29-Jan-2010 23:02 | BrianH | Comment : 0001914 | Modified | - |
| 24-Jan-2010 05:36 | BrianH | Comment : 0001914 | Modified | - |
| 24-Jan-2010 04:13 | BrianH | Comment : 0001914 | Added | - |
| 20-Jan-2010 23:52 | PeterWood | Comment : 0001911 | Added | - |
| 20-Jan-2010 23:45 | PeterWood | Comment : 0001910 | Added | - |
| 20-Jan-2010 15:48 | abolka | Comment : 0001909 | Added | - |
| 20-Jan-2010 12:29 | pekr | Comment : 0001908 | Added | - |
| 20-Jan-2010 08:16 | PeterWood | Ticket | Added | - |