[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: Shared MIME-Info 0.10



On Thu, Mar 06, 2003 at 11:54:52AM -0500, Jonathan Blandford wrote:
> Waldo Bastian <bastian kde org> writes:
> > 
> > As long as such formalisation takes into account that the content matching 
> > step can not always be performed for a variety of reasons.
> 
> So you mean proposing a 'slow' method and a 'quick, approximate' method?
> Hmm.  I could definitely live with that.  Might be good to make sure
> that cutting out the magic step does actually make it significantly
> faster (which I assume it will.)

There are several different forms of matching

1) match given only a name
    File managers and dialogs that need to sniff lots of files
    quickly would want this as a firt pass approximation, or as the
    only step for network files.

    Also Useful when guessing what format to save in given only a
    filename (eg user says save as foo.xls and I guess Ms Excel)
    Certainly not the core use case, but still relevant.

2) Basic content matching
    Use file magic type sniffing to distinguish things reasonably
    well.  Its still not going to catch some of the more generic
    formats (eg its a zip file) but it gives you a start.  File
    Managers may use this is a validation step after the name globs.

3) Expensive content matching to do things like
    - match on sub-stream names (MS Excel vs Quatro Pro or OO vs zip)
    - match on xml namespaces (dia vs sodipodi)
    - transparent decompression





[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]