Persistence(New Deal), pointers&Linbot(site maintenance)

[Billy Tanksley]

> From:	dufrp@oricom.ca [SMTP:dufrp@oricom.ca]
> Subject:	Persistence(New Deal), pointers&Linbot(site maintenance)
> 
> Because New Deal is also orthogonally persistent?
> 
	I don't know what you mean.  "Because" is a connective -- what are
you claiming is caused by orthagonal persistance?

> The demo of New Deal impressed me about 1 year ago,
> truely it is able to do some nice stuff with few
> ressources. And if you tell me it is persistent, I'll be
> even more impressed.
> 
	It is.

> However that's commercial stuff and
> this alone is enough to let me away of it.
> 
	Understood.  It's a good company, but it's not the same as free
software.

> But how does it influence the system, is persistent bug a
> reality, does menu come with a "don't save" option? :-)
> 
	System state is only saved on shutdown, so only a successful
shutdown will be permanant.  Document state is persistant, but is saved in
three places: the "saved" document (only saved when the user requests it),
the persistant store (in the same file as the saved document, stored as a
difference form the last save), and the backup document (a seperate file in
a seperate directory created fresh whenever the user clicks the backup
button).

	The concern about keeping the user's data safe is one thing which
really impresses me about Geos.

	Compare this with OS/2, which attempts to restart all applications
which were running when it was halted without a proper shutdown.  The
problem is, the most common reason for lack of a shutdown is that one of the
apps started hogging 100% resources and the system became unusable.  So now
OS/2 decides to start that app up again...  Sigh.

	The really sad part is that there's no good use for OS/2's sorry
attempt at persistance.

> That's a long time ago I ask myself how GC and persistent
> do treat pointers. I means you have to know what is a
> pointer. In a Lisp language it is quite easy to know what
> is a pointer but in a OS, how do you follow the pointers?
> One way to achieve that that come to my mind is to have a
> pool of pointers and each time an application want to
> allocate something return it a pointer in the pool.
> That's quite the way Mac have done it since the beginning
> if I am right. But I feel there is a simpler or better
> way, is it?
> 
	That's also how Geos does it.  I feel that there's a better way, but
I don't know what it is.

> Well, Linux Debian is slowly installing itself, I love
> the Lynx browser, gcc is working, mail is hard to
> configure (said to exim I was a satellite, and still need
> to learn to use fetchmail), have cvs, need to learn to
> use it (have curves to help me), have python, forth, etc.
> Now I have seen the Linbot package (search 'linbot' with
> / in dselect), it looks like something that could help
> maintain the web pages, that is find the broken links
> daily if you want and give an html report.
> 
	Debian rules.

	-Billy