Re: Request: Test suite for EFS.



On Thu, Feb 17, 2000 at 06:54:56PM -0600, Miguel de Icaza wrote:
> 
> > > Can you picture a way in which this could be solved with XML?
> > 
> >   Not just XML, but a combination.
> > Server runs WebDAV, all WebDAV servers allows to associate properties
> > to a given resource. Maciej having edited that document is such a metadata
> > which should be stored on the server. Those metadata could be accessed
> > without even opening the resource in question. Indexing them and maintaining
> > a complete coherent view on the server is far easier taht way.
> 
> This sounds like something few people will ever get right.  I was
> thinking "NFS file sharing" here.  And still then, you do need a
> pretty good WebDAV server for this.

  mod_dav Apache module from Greg Stein provide the properties mechanism.
However there is no indexing as far as I can tell. This work is under
the DASL (spelling ?) WebDAV searching IETF working group. I understand they
are making progresses. Linking to LDAP databases is also possible.
  It's probably not 100% normalized but the basic building blocks are there
available in opensource products, and it doesn't sound like rocket or
AI science to me. One need to spend hacking time on it to build a solution
though, that's right !

> > > The entire scenario described above is avoided completely in current
> > > Microsoft environments, because they can just scan the documents for
> > > the "summary/author" inside each file.  Does not take a lot of memory,
> > > and does considerably less IO.
> > 
> >   Current environments, maybe, future, I doubt. 
> 
> Sigh.  When we do have infinite memory and infinite CPU speed, it
> might not be an issue.  What about the next 2 years?

  I do think taht opening each and every binary package to find the information
located in it might take far more CPU and memory than querying an adequately
filled LDAP database for the same pattern. Maybe I'm wrong.

  Those metadata information about the document stored on the server
can also be serialized as a separate file (in XML format or whatever
else if you get allergic to XML) with reference to the components on the
server, or on the local directory if you save to a local filesystem 
like on a floppy. You can then use that metadata file to build 
a zip/tar based format (c.f. the Jar format for java packages)
if your *really* need a single resource.

   Save  -> write back to the file system and update the metadata
   SaveAs -> if on an FS without metadata write the individual components
             and the metadata file 
   SaveAzZip -> write a Jar like file.

Doesn't sound like something for Y2010.

> >   Well i would rather spend time reimplementing new MS technologies
> > like XML based stuff, WebDAV, etc. rather than spending time on the
> > older ones especially if one knows it's limitation are barely
> > acceptable.
> 
> I have yet to see why structured storage is "barely acceptable".  

  As you told OLE serialization has for example 32 byte max name lenght.
and this is a serious limitation in your mind I understand. EFS has the
very serious limitation of being a new binary format usable only by 
the Gnome tools compiled with the right bonobo stuff, I consider that
a barely acceptable limitation. You can argue this point though.

> >   Well that's not a good point in their favor. Do not follow them
> > on that.
> >   Maybe a better handling of multiple components documents can be
> > done later as an alternate serialization mechanism (XML based or not).
> > Doing it right the first time is nearly impossible. But let's learn
> > from others (and especailly MS's) mistakes.
> 
> I am all for learning from other people's mistakes.  But fact is: they
> do have a lot of experience in this field (whoever says "no they dont"
> needs a RealityCheck(tm) pill).
> 
> So, how exactly is Microsoft's approach for this problem wrong?

  Depend what MS division >:-> this is a very large company ...
OLE serialization has limitation (but would be a good solution if you
want to "sell" gnome apps to MS users).

Daniel

-- 
Daniel.Veillard@w3.org | W3C, INRIA Rhone-Alpes  | Today's Bookmarks :
Tel: +33 476 615 257  | 655, avenue de l'Europe | Linux XML libxml WWW
Fax: +33 476 615 207  | 38330 Montbonnot FRANCE | Gnome rpm2html rpmfind
 http://www.w3.org/People/all#veillard%40w3.org  | RPM badminton Kaffe



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