[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
rtems-980923 is available
- Date: Wed, 23 Sep 1998 17:43:58 -0400 (EDT)
- From: qqi at world.std.com (Quality Quorum)
- Subject: rtems-980923 is available
On Wed, 23 Sep 1998, Quality Quorum wrote:
I am responding to my own mail, after giving it some thought, and
reading other mail.
Precise exresion is 'return value is a number of characters
yet to be dequeued'.
IMHO, returning 0 and 1 is very unintuitive.
BTW, I think that you guys are also wrong on the matter of enforcing
common style, let us not discuss both these issues, the thing will
work either way. I am just appalled by the level of nitpicking I saw
> On Wed, 23 Sep 1998, D. V. Henkel-Wallace wrote:
> > At 10:02 23-09-98 -0600, Eric Norum wrote:
> > >Perhaps the comment at the beginning of
> > >rtems_termios_dequeue_characters() should be modified to note that
> > >the return value is `the number of characters that remain to be
> > >dequeued, although some of those characters may be in the process of
> > >being transmitted'.
> > I have no comment on the specific technical issue, but this suggestion I
> > clearly don't like.
> > If the return value is of unpredictable accuracy, then responsible
> > programmers shouldn't use it. But commenting it lends it an air of
> > legitimacy that could lead less experienced programmers to use it.
> > Although this is a small case, it's basically the equivalent of leaving a
> > loaded firearm lying around.
> > In a case like this it's better to resolve the "does so"/"does not" issue
> > than to finesse it with a compromise.
> > If this value is really so unstable (equivalent to choosing a random number
> > between 0 and the input count) it's hard to believe that it could make
> > another function cleaner to implement.
> > Cheers,
> > g
> > PS: hope you don't mind my message Alexy -- I appreciate your patches too!
> I am sorry, right comment should be that it is value used in
> last call to driver write function or 0. This value is used in 16550 and
> similar serial drivers to indicate whether or not to keep TX interrupt
> enabled, I was thinking about returning 0 and 1, I decided against
> it because it, may be I should not.