[bind10-dev] [bikeshed] "variable == value" vs "value == variable"

[Kevin Tes]

> In review of recent tickets I've noticed some developers seem to
> prefer this style:
> (Style A)
>    int x;
>    ...
>    if (10 == x) {
>        // then do something that has to be done when x is 10
>    }
> 
Yeah,Jinmei this style we CNNIC members prefer. The reason is your "In
my understanding" section.

Here we intend to develop codes which can be easily or simple catch by
complier if some 'syntax' or mistype errors exit 'in case the complier
is not smart enough', so this style was used.

> I'd propose we write instead
> (Style B)
>    int x;
>    ...
>    if (x == 10) {
>        // then do something that has to be done when x is 10
>    }
> 
> and add it to our style guideline.
> 
> My reason for the suggestion is readability and style consistency.
> 
> Regarding readability, Style A reads quite awkward to me.  Style B can
> naturally be understood as "If x is equal to 10, then...", but Style
> A requires some transformation inside your brain from a C/C++
> expression to the human-understandable semantics.
> 
> In my understanding, the only benefit of Style A is that it could be
> more robust against a typo like "10 = x" (which is a syntax error and
> can be detected at compile time, while "x = 10" is a valid
> expression).  But I'd say this defence doesn't outweigh degrading the
> readability in the world with modern compiler/language.  Modern
> compilers generally warn if we omit one '=' in style B:
> 
> foo.cc:18: warning: suggest parentheses around assignment used as truth value
> 
> (with some higher level of warning options), and with something like
> -Werror we can still detect it at compile time.
> 
> Also, in C++ many more variables can (and should) be const, and if the
> variable is const we can even detect the typo at compile time based on
> the language specification.  If variable x in Style B is const and we
> omit '=' in the if condition, we'll see something like this:
> 
> error: assignment of read-only variable 'x'
> 
> In my opinion, we should rely on computers and compilers (and
> sometimes the power of language rules) for avoiding and detecting
> human errors like this, rather than by introducing less understandable
> convention.

If you think it is less understandable, it is ok to add this style to
the codes guide line.
> 
> As for consistency, this is probably minor compared to inconsistency
> on (e.g.) the indentation width.  But to me the mixture of these two
> styles still imposes the impression that the code isn't well
> maintained, allowing arbitrary changes to be introduced.  So at the
> very least I'd like to use (only) either of the styles consistently -
> although I strongly prefer Style B, if the majority of the developers
> insist Style A is better, I'd rather suggest using that style
> throughout our code.
> 
> ---
> JINMEI, Tatuya
> _______________________________________________
> bind10-dev mailing list
> bind10-dev at lists.isc.org
> https://lists.isc.org/mailman/listinfo/bind10-dev
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 316 bytes
Desc: This is a digitally signed message part
URL: <https://lists.isc.org/pipermail/bind10-dev/attachments/20120325/7b139704/attachment.bin>