--- Christian Gross <[EMAIL PROTECTED]> wrote:
Ok so APR doesn't have thread pools? Could this be implemented in the future or doesn't the existing architecture allow for that?
APR is not locked into specific models. It could be added on. (But others may correct me if I am wrong). That is part of the difference with APR and NSPR. APR can easily be "fixed" whereas NSPR has problems with this.
>Ok I may be biased but I prefer APR. I have used both extensively >and find that APR is just simpler to use and more productive.
Thats certainly an important point.
It is and this is what really convinced me of APR.
>For Apache http://www.devspace.com/Technology/ApacheStuff.html and >for NSPR http://www.devspace.com/Technology/MozillaStuff.html.
There doesn't seem to be anything on the MozillaStuff page?
Yeah the NSPR demos. Look at the end of page. BTW look forward to the end of this month for more APR stuff. I am right now in the process of finishing cServer, which is a C++ wrapper that uses pools natively for memory allocations.
>Part of the problem with NSPR is that its build process is too darn >complicated and integrating NSPR is not much fun. Another problem >with NSPR is that a few times I have had conflicts with naming of >NSPR data types. This has never occured with APR.
Cool. Thing is, I'm toying with the idea of writing an OO DBMS. So scalability and high performance are extremely important. Oracle uses a portable runtime, btw, there's is called VOS (Virtual OS).
I would REALLY recommend APR!!! APR is simply better suited for that. Right now I am writing a JavaScript 2.0 implementation for JSModeler and APR saves my bacon!!!
Christian Gross
