On 29 Dec 2001 20:09:59 +0100
Erik Bågfors <erik_at_bagfors.nu> wrote:
> On Sat, 2001-12-29 at 20:04, Alan Cox wrote:
[...]
> > Doesn't help you. What if the other app isnt running.
> You have to start it ofcourse. That's what's bonobo is doing for
> example. That's why I wrote that I was stupid because that's probably
> not a good idea on a device with limitied hardware..
Yep...
I hope that the application startup will be pretty quick when using C for
the main applications but even then it is undesirable to start every
application you need data from.
> > > addresslist. If I just add that to the other address-application's
> > > database and that program is running it will not know that I added
> it.
> > SQL has extensions for this too
> Really?? Does SQLite support this??
Good point ;)
Let's have a look at it!
> > > On the other hand maybe this will leed to a situation where all
> > > applications are running at all times and that's not good at all on
> a
> > > small device with limited memory. =20
> > Right. I'd rather the database (be it SQL or otherwise) kept the one
> > cached copy of the data, used its knowledge that it is the only writer
> to
> > buffer bursts of overlapping changes so that we reduce flash writes,
> and
> > handled the notification of changes.
> I didn't even think about flash writes.
This is very important though I think we have to look a little bit closer
to it.
I think we can assume that we will use Familiar as underlaying
distribution. Using Familiar we get JFFS2 as filesystem which is supposed
to handle the flash writing part pretty well, i.e. compress the file and
sequentially fill the flash before rewriting it. On top of this we might
come up with a double storage mechanism like storing in RAM and doing a
flash copy just once a day to save some flash cycles.
But even then you still have the problem how to handle record update or
delete within a XML file.
> Ok. Maybe I'm getting convinced. I just think it's very easy to use
> XML but it might not be a good idea on a small device like this.
Hey, don't be disappointed, OK?
This discussion is very interesting and the more the others like me have
to defend their position the clearer it becomes.
So please keep up questioning (and while we are at it, there are _no_
stupid questions ;)
> /Erik
CU
nils faerber
-- kernel concepts Tel: +49-271-771091-12 Dreisbachstr. 24 Fax: +49-271-771091-19 D-57250 Netphen D1 : +49-170-2729106 --Received on Sat Dec 29 2001 - 11:23:43 EST
This archive was generated by hypermail 2.2.0 : Mon Jul 25 2005 - 17:18:53 EDT