# The following was supposedly scribed by
# André Warnier
# on Tuesday 15 March 2005 05:26 pm:
>On the other hand, mod_perl seems to have some deep interactions with
>the innards of the Apache server, for example direct manipulation of
> the "bucket brigades". How come no-one has come up yet with some
> scheme to "hijack" one of these bucket brigades (or some other
> internal Apache memory area) to provide such a capability ?
>
>Also, for instance, the DBI module and it's companion Apache::DBI
>provide persistent connections to databases. And to my knowledge, they
>are available under Unix as well as Windows. For doing this, there
> must be some form of memory sharing at work, isn't it so ?
The answer to these and all related "why" questions probably contains
the word "security".
Any kind of memory sharing between apache processes is a potential hole,
so you have to be really careful about how you code it.
I think that what you propose is certainly possible, but it would be a
big task to cover all of the vulnerabilities. So, your best bet is to
use something that already exists, and there's been a lot of work done
by a lot of people on the DBI, so that's probably where you'll get the
most mileage.
You keep saying that you want to share memory, but what you really want
to do is share data. In terms of security, stability, etc. you solve a
lot of problems by avoiding shared memory. With that in mind, try to
reframe your problem and you'll probably find that DBI will be the path
of least resistance. There are a few wrapper modules that simplify the
interface (some even tied to hashes?) so maybe you want to use one of
those if your frustrated by having to speak in SQL to talk to your data
structure.
--Eric
--
"Left to themselves, things tend to go from bad to worse."
-- Murphy's Corollary
---------------------------------------------
http://scratchcomputing.com
---------------------------------------------
