Hello all,
I was going through the ingres code, and found this comment:
/* Give up if unable to get memory in one minute */

Arrrrgggg!!!!
Time dependencies!!!!
Is this sort of stuff that you would expect to find in production
quality code?

Plus, I would like to know if there are any obvious advantages to using
process oriented databases like PostgreSQL and Ingres over ones that
use threads(IMHO much cleaner) apart from the fact that they are more
stable -- if a process crashes, it doesn't take down everything with
it.
Practically, software does have bugs, but assume for the time being
that databases are bug free.

Regards,
-Dhruv.

"dhruv" <dhruvbird (AT) gmail (DOT) com> wrote

Is there a constructive reason why you cross posted that "question" to
non-Ingres groups? Since you don't claim to work for an enterprise that
has actually been adversely affected by whatever this comment refers to, I
presume you are just a troll. But you might persuade me you are genuine if
you try.

Roy

Roy Hann wrote:
Yes, to know what others have to say about it. If you notice,
PostgreSQL is also process based -- fork()ing on a new connection,
while MySql is thread based. It's generally good to know the opinion of
people who have worked with systems having varying designs.

Regards,
-Dhruv.

On 3 Nov 2006 03:10:56 -0800, "dhruv" <dhruvbird (AT) gmail (DOT) com>
wrote:

Most definately yes. Your operating system and network stack are
full of time dependencies. It makes no sense to wait forever for
a resource you can't get.

You don't want any service to give up immediately when some
resource is temporarily unavailable. That would be a nuisance to
any application using it.

At the same time you do want your application to tell you when
and why it couldn't accomplish its task, instead of silently
being stuck, holding other resources, until you kill it....

To reach that goal the layer below your application has to give
up after a reasonable amount of time, and return with an
errorcode, or throw an exception. It also may have to give up to
avoid or resolve deadlocks.
One minute is a long wait for a local resource like memory

I left comp.databases.mysql out, it's OT there.
Fup-To set to comp.databases.ingres,comp.databases.postgresql

HTH
--
( Kees
)
c[_] If it wasn't for C, we'd all be using BASI, PASAL and OBOL (#105)