On Tuesday, October 19, 2021 12:59:56 PM CEST Karolina Drobnik wrote:
> Hi,
>
> Thank you very much for your comments.
>
> On Mon, 2021-10-18 at 17:12 +0200, Greg KH wrote:
> > Also, these are all just fine as-is for now. A better way to make
> > these lines smaller is to use better variable and function names
> > that are shorter and make sense :)
>
> I have v2 ready but I'm not sure, given the Joe's patch, if my solution
> is a satisfactory one. I didn't jump on such refactoring as I'm still
> learning about the codebase/process and didn't want to muddle the
> waters (...more than I do already).
>
> Greg, what would you prefer? Should I back up with my patch, pick
> something else and let Joe's patch be merged?
>
>
> Also, I have a question about the patch if that's ok :)
>
> On Mon, 2021-10-18 at 22:56 -0700, Joe Perches wrote:
> > Maybe some refactoring like:
> > ---
> > drivers/staging/vt6655/rf.c | 38
> > ++++++++++++++++++--------------------
> > 1 file changed, 18 insertions(+), 20 deletions(-)
> >
> > diff --git a/drivers/staging/vt6655/rf.c
> > b/drivers/staging/vt6655/rf.c
> > index 0dae593c6944f..7beb0cd5a62df 100644
> > --- a/drivers/staging/vt6655/rf.c
> > +++ b/drivers/staging/vt6655/rf.c
> > @@ -680,16 +680,19 @@ bool RFvWriteWakeProgSyn(struct vnt_private
> > *priv, unsigned char byRFType,
> > u16 uChannel)
> > {
> > void __iomem *iobase = priv->PortOffset;
> > - int ii;
> > + int i;
> > + unsigned short idx = MISCFIFO_SYNDATA_IDX;
> > unsigned char byInitCount = 0;
> > unsigned char bySleepCount = 0;
> > + const unsigned long *data;
> >
> > + uChannel--;
> > VNSvOutPortW(iobase + MAC_REG_MISCFFNDEX, 0);
>
> I see that you introduced `uChannel--` to further tidy up the lines
> with `[uChannel - 1]`. In general, is there anything wrong with
> indexing like `i - 1`? What's the preference here? DRY things up as
> much as possible?
Hi Karolina,
No, there is no problem in using a[i - 1]. Personally I prefer the former
when 1 <= index <= ARRAY_SIZE(a). If you code "index = index -1;" or
"index--;" (that is the same) and then you use 'index' many lines below that
decrement in "a[index]" it may be not immediately clear that you are not
indexing past the end of the array.
But this is not the point. You may still use Joe's style or leave it as
"index -1". The point is that Joe is just showing you some different way that
you can use to accomplish the task of "Fix line wrapping in rf.c file".
He put many different changes in one only single patch. Maybe that those kind
of patch are permitted to developers who have gained so much trust that Greg
doesn't need anymore to check "one {,logical} change at a time" (but still
I'm not sure about it).
I guess that if Linus T. or Greg K-H. want to put many different things in
one big "Clean up rf.c" patch they can. This (yours) is not the case. If you
decide to use one or more of the example Joe showed you you must be careful
to split changes in a series of patch, according to the instructions you read
in the Outreachy pages at kernelnewbies.org.
Joe is showing that you can shorten lines with several techniques...
1) renaming variables:
("ii" => "i") and getting rid of hungarian notation ("bySomething"
=> "something");
2) contracting instructions:
"uChannel--" or "*data++" - for the latter take care of preceding
rules or, better, use redundant parenthesis like in "*(data++)" to
facilitate readability and comprehensibility);
3) using temporary variables:
"unsigned short idx = MISCFIFO_SYNDATA_IDX;" or "const unsigned
long *data = dwAL2230InitTable;");
4) refactoring lines of code (e.g.,
- if (uChannel <= CB_MAX_CHANNEL_24G) {
- for (ii = 0; ii < CB_AL7230_INIT_SEQ; ii+
+)
- MACvSetMISCFifo(priv, (unsigned
short)(MISCFIFO_SYNDATA_IDX + ii),
dwAL7230InitTable[ii]);
- } else {
- for (ii = 0; ii < CB_AL7230_INIT_SEQ; ii+
+)
- MACvSetMISCFifo(priv, (unsigned
short)(MISCFIFO_SYNDATA_IDX + ii),
dwAL7230InitTableAMode[ii]);
- }
+ data = (uChannel < CB_MAX_CHANNEL_24G) ?
+ dwAL7230InitTable :
dwAL7230InitTableAMode;
+ for (i = 0; i < CB_AL7230_INIT_SEQ; i++)
+ MACvSetMISCFifo(priv, idx++, *data++);
5) something else that I'm missing and that you may easily notice :)
I prefer to state it again: if you choose to do such kind of works, be
careful to split self-contained patches in a series and explain each change
you make and why you make it. Each patch must do only one logical change.
Each patch of a series must be self-contained also in the sense that it must
build without introducing errors or warnings at any point: for instance, five
patches => five clean builds.
Thanks,
Fabio M. De Francesco
>
> I'm asking because when I was reading this line, at first, it wasn't
> clear to me why we could decrement it (example though: "Was this
> modified earlier? Do we need to "correct" it?").
>
>
> Thanks,
> Karolina
>
>
> --
> You received this message because you are subscribed to the Google Groups
"outreachy-kernel" group.
> To unsubscribe from this group and stop receiving emails from it, send an
email to outreachy-kernel+unsubscribe@xxxxxxxxxxxxxxxx.
> To view this discussion on the web visit https://groups.google.com/d/msgid/
outreachy-kernel/810a4e29b0c54520a30cae4d37fde0a59ea3d83b.camel%40gmail.com.
>
