[openbeos] Re: Development Model?
- From: Scott Mansfield <thephantom@xxxxxxx>
- To: openbeos@xxxxxxxxxxxxx
- Date: Tue, 13 May 2003 18:45:38 -0700
Hi Waldemar,
On Monday, May 12, 2003, at 10:29 America/Los_Angeles, Waldemar
Kornewald wrote:
What are you thinking of when you want to "change aspects of
BNetEndpoint"?
As this message was already posted on every possible group I am
willing to
talk about it, but I have not understood your problem. ;)
I only posted to the openbeos list because, although I am working on
BNetEndpoint, I felt the problem domain was not specific to the
networking group. How it wound up on "every possible group" is beyond
me.
[-=Snip=-]
Do you mean that InitCheck() only returns B_ERROR instead of the real
error?
Please excuse my stupidity. :)
It's not your stupidity. The stupidity is mine alone, curse my
flu-addled brain. One should never take Nyquil and then attempt to
code or write coherent e-mails. :-)
Anyways, I was struggling with some vagueness in the BeBook. In the
BNetEndpoint class the Socket accessor returns member data (an int,
from a call to ::socket()). The BeBook states that this accessor
"Returns the actual socket used by the BNetEndpoint for data
communications." Here are my concerns:
1. What if a particular instance is not initialized, do I return
B_NO_INIT (or B_ERROR)? These are not sockets. Yes, I know that
BNetEndpoint's consumer is supposed to call "InitCheck()" and act
accordingly before doing anything else -- but we're not living in a
perfect world; I've seen way too much bad windoze code to know
otherwise.
2. We return a non-const 'fd' to an opened socket, free for our caller
to trounce upon willy-nilly. Our caller can even close( socket ) on
us, leaving whatever instance of BNetEndpoint 'socket' came from in an
indeterminate state.
Thanks for your time.
Cheers,
Scott
Other related posts: