[no subject]

   MySQL supports two different kind of tables. Transactions
   safe tables (BDB) and not transaction safe tables (ISAM,MyISAM and
   HEAP.

   Berkeley DB (http://www.sleepycat.com has provided MySQL with a
   transaction safe table handler. This will survive crashes and also
   provides commit and rollaback on transactions.

<snip>

   If an update fails, all your changes will be restored. 

<snip>

   Even if Berkeley DB is in itself very tested and reliably, the MySQL
   interface is still very alpha, but we are actively improving and
   optimizing it to get it this stable real soon.

<snip>
 
-----------

There's a list of statements not supported by the BDB tables yet. 

All in the MySQL manual. 
 

> 
> Read: http://openacs.org/why-not-mysql.html , but take it with a grain
> of salt. 

People who opt for MySQL are aware of what's lacking in MySQL when they
make that decision. I guess performace is required instead of transaction
support in most cases.

That article was probably written to get some publicity for the arsdigita
community system which no one wants to use :>
  
> > 
> > Do Free/Net/Open BSD have a journalling filesystem?
> 
> No but they support softupdates, which became free last week, previously
> it was available, but for non-commercial use only.
> 
> There is an accompanying USENIX paper, also presented last week,
> which tries to argue that softupdates might be a better approach than
> journalling. See my posts in the last week for the link.
> 
> The difference is - do you want to rollback the transaction and go to
> a consistent state (this is the approach journalling takes). It may
> make sense for databases. Does it make sense for filesystems ?
> 
> Softupdates doesn't try to do transactions, but ensures that the 
> filesystem is always in a usable state. Example: when you rename 
> file a to b, journalling ensures that you see file a or b, but not
> both. Softupdates might result in having both a and b. But do you
> really care ?

hmm, not really, some apps may but I guess they have to be robust enough
to deal with that.

> 
> They both achieve the same results (short fsck times), but have different
> overheads.

That's interesting. Will check it out when I finish downloading OpenBSD
and find a machine to put it on.

Now, is there softupdates for Linux? 

					- cnb.