Joe Eykholt wrote:
> Robert Love wrote:
>> On Mon, 2009-06-01 at 14:28 -0700, Joe Eykholt wrote:
>>> Robert Love wrote:
>>>> On Mon, 2009-06-01 at 11:10 -0700, Joe Eykholt wrote:
>> 
>> <snip>
>> 
>>>>> Along with this, I'd like to rename struct fc_rport_libfc_priv to
>>>>> something else. Any suggestions?  How about fc_peer?  or
>>>>> fc_pport?  I want something short like that, not specifically a
>>>>> target or initiator, not necessarily an N port or 
>>>>> a discovered port, maybe just a port?
>>>>> 
>>>> Of the ones you list I think I like fc_pport best, although it's a
>>>> bit strange. fc_peer doesn't mean that much to me since it's
>>>> missing "port" and fc_port would falsely indicate some type of
>>>> inheritance (that fc_lport should be a child of fc_port) or
>>>> something (to me at least). 
>>> I have the same misgivings, and after starting to code this up, it
>>> seems really strange to have a function called fc_rport_login that
>>> takes a pport as an arg, ... that would make us want to call it
>>> fc_pport_login, and then to rename fc_rport.c to fc_pport.c ...etc.
>>> ... it get's too messy. 
>>> 
>>> Also, despite the name, fnic is accessing fields in
>>> fc_rport_libfc_priv, to find out whether to do retry, etc., on
>>> offloaded ops. 
>>> 
>>> So, to keep things sane, I'd like to keep the name the same, or
>>> maybe just drop the libfc portion. (fc_rport_priv).  Less code
>>> changes. 
>>> 
>> 
>> What do you think of fc_rport_libfc, or fc_libfc_rport? I think I
>> like the prior better. It replaces "priv," which doesn't have much
>> meaning to "libfc," which does. 
>> 
>> Either way it's not a terribly big deal to me and I do prefer
>> fc_rport_priv to the initial alternatives.
> 
> I think having the module name in the structure name isn't so nice.
> I guess it's more descriptive than priv, since it says who it
> belongs to ... maybe fcs_rport?
> 
> Or, how about fc_login, since it describes the state of the login?
> That's what fc_sess was trying to do, I guess.  But I think it should
> start with fc_rport_ so we don't feel the need to change the name of
> the file and all the function names that start with fc_rport.  That's
> why I called it fc_rport_priv.
> 
> One thing that sort of bugs me about this is the use of rdata all
> over the 
> place instead of rport, as the variable name that's by convention
> used for fc_rport_priv.  Maybe the variable name should be rportp or
> rp. 
> 
This was something discussed internally, at least, durring the re-arch.
The debate at the time was "lp" vs. "lport" and "rp" vs. "rport". I
think "lport" and "rport" were used since the FC transport was using
"rport" for members. I believe the "rdata" was chosen because another
FC driver was using that name for it's private data.

"rp" and "lp" aren't any more grep-able than their longer counterparts.
I don't care as long as we're consistent.

> It's a simple patch to change the name all over the place, and I don't
> mind doing that at some point.  For now, I'll stick with
> fc_rport_priv and rdata. 
> 
>       Thanks,
>       Joe

_______________________________________________
devel mailing list
[email protected]
http://www.open-fcoe.org/mailman/listinfo/devel

Reply via email to