Re: [e-lang] File API taming
"Stiegler, Marc D" <[email protected]>
| Newsgroups | gmane.comp.capabilities.general,gmane.comp.lang.e.general |
|---|---|
| Message-ID | <DED48AFB6A17C5469D692D2CB60A4AB51D96276D4D@GVW1155EXB.americas.hpqcorp.net> |
> If you view the name of the file only as a petname within the > namespace defined for the containing directory, sharing the > value of this petname without sharing a capability to the > containing directory would not transfer any authority, > especialy if authority to the file itself is shared already. > > Viewing names like this however does not make the name a > property of the file object, but a property of the relation. > You may use this concept to create some sort of 'relation' > proxy object that transparently proxies the file object and > extends the interface with a getName() method. > > Extending this concept to a getPath() implementation however > seems to give rise to possible confusions. People are used to > using absolute paths for designation, thus providing a full > path with an artificial getPath() implementation may lead > people to mistakingly look for ways to use these paths as > designations. > > Rob There are 2 ways they might think of such a path as a designation, one correct and one incorrect. The incorrect way would be to think of it as a designation that bears authority. but that mistaken usage is quickly self-correcting, because one immediately notices one has no way of converting the string to a power. The other way to use this as a designation is as a richer description for coordination among users who already have the authority and are trying to figure out which of their similar authorities they are about to use together. This is a correct use. And even though this use will be important rarely as a percentage of uses, given millions of users, it will be important often. --marcs