Re: Request: Test suite for EFS
- From: Kent Schumacher <kent structural-wood com>
- To: gnome-components-list gnome org, gnome-list gnome org
- Subject: Re: Request: Test suite for EFS
- Date: Wed, 16 Feb 2000 11:53:45 -0600
- From: Kent Schumacher <kent structural-wood com>
- To: Miguel de Icaza <miguel gnu org>
- Subject: Re: Request: Test suite for EFS.
- Date: Wed, 16 Feb 2000 08:03:46 -0600
Miguel de Icaza wrote:
> > Is that the end of the XML format ?
> > How are you gonna reuse your filesystem embedded in a file 10 years from
> > now when you will look at your precious data on a CDROM but without a running
> > copy of Gnumeric ?
> >
> > Daniel, concerned ...
>
> It is definetly an option.
>
> First of all, some features are very hard to implement with ad-hoc XML
> features. For example, it is not easy to write an application that
> scans documents and searches for "Daniel Veillard" as their author.
>
> With the way Structured Storage works in OLE2, you know that the
> author is always going to be in "summary/author" inside the structured
> storage file: there is no need to load the entire file, you just open
> the directory, and do the lookups for summary/author inside the file
> system, no need to parse the entire thing into memory and find out
> whether the author is Daniel or not.
>
> Inside this structured storage file, the current XML file will be
> contained as well as any other resources like images, links and
> embedded objects.
>
> Unless someone convinces me that this is not a good idea, Structured
> Storage makes a lot of sense.
>
> And of course, we could support a "pure" XML mode for those that do
> not want to use the search facilities.
>
> miguel.
Could there be such a thing as a compiled indexed XML file that could
be compiled (or decompiled) using separate programs (much like tar
operates).
Programs could load specific tagged items via a call that supplies
an arg list of tags to be retrieved. The call would have the responsiblity
of determining if the file was XML or compiled and behave appropriately.
Then you could get the efficiency of binary indexed files and not give
up the ubiquity of XML files. The user could choose which he preferred
on a file by file basis.
Kent
[
Date Prev][
Date Next] [
Thread Prev][
Thread Next]
[
Thread Index]
[
Date Index]
[
Author Index]