Re: Moving away from app-centric mimetypes (e.g. kword)
Sven Niedner <[email protected]>
| Newsgroups | gmane.comp.kde.look |
|---|---|
| Message-ID | <[email protected]> |
On Wednesday 15 May 2002 02:15, Steven D'Aprano wrote: > Am I off-case here? If what I say is unreasonable, I'd like to hear > your feed-back. I agree with you. However, I again advocate for a database driven solution. Some poster reminded of the "store files in cvs" discussion, and other suggested a database. There is acutal need for these features. Another poster did not speak of file access, but retrieval, which is more general. My suggestion is again to design an abstraction layer based on requirements: - version management - meta data handling - access rights - directory structures - etc. Once an abstraction layer is completely desinged, the backend can be discussed: How can these requirements be mapped to: - local file systems w/o. meta data support - local file systems with meta data support - remote (SMB, NFS, FTP, etc.) filesystems - Databases - CVS - other kioslaves - etc. This way we can use the same metadata management tools throughout all environments; in a coroporate environment, files would most likely be stored in a database. In a developer-centric environment, most likely CVS, and so on. An application programmer does not need to care about it, he can just acces the file load/save methods. This approach is powerfull for the ioslaves and should be used again. Another useful thing would be storing the history of a file: you write a letter to X using a template. Another day, you load it,, modify it, and save it under a different name. This letter should remember its parent(s), and the initial template. Meta data should be inherited, too. Greetings, Sven. -- I don't have an email address. I reached an age where my main purpose is not to receive messages. (Umberto Eco)