[prev in list] [next in list] [prev in thread] [next in thread] 

List:       kde-pim
Subject:    Re: KDE PIM Roadmap (Call to action)
From:       Rik Hemsley <rik () kde ! org>
Date:       2001-06-16 8:55:06
[Download RAW message or body]

#if Guillaume Laurent
> On Saturday 16 June 2001 03:12, Rik Hemsley wrote:
> 
> > And it's not hard to corrupt a text file ?
> 
> No, but it's much easier to recover from a parsing error, and also repair it.

I meant a single file as opposed to multiple files. Certainly text
is just fine for storage of this type, as we're not expecting to
host huge quantities of data. I also prefer using text (as opposed
to binary) myself.

I've use XML in the basic kab2 backend. This removes the need for yet
another hand-written parser and also keeps readability.

Of course, using binary is possible too, simply by using
operator << (QDataStream, KAB::Entry)
instead of
KAB::Entry::insertInDomTree(QDomNode, QDomDocument)

Not that I think anyone will want to do that.


> > For kab2, I've used one-file-per-record. It's fast and safe.
> 
> That's also a possibility.

Benefits:

* No locks needed - safe over NFS and can be accessed by multiple
  processes if desired.

* Safe - an application or system crash can't corrupt data.

* Just as fast as a flat file.

Drawbacks:

* One inode per record -> some 'wasted' space. But then storing
  in plain text isn't ultra-space-efficient anyway.

Rik

[prev in list] [next in list] [prev in thread] [next in thread]