Bobby Holley writes:

> On Wed, Jul 8, 2015 at 3:52 AM, Gabor Krizsanits <gkrizsan...@mozilla.com>
> wrote:
>
>> The priority is to automatically rewrite our source with a unified style.
>>> foo -> aFoo is reasonably safe, whereas aFoo->foo is not, at least with
>>> the
>>> current tools. So we either need to combine the rewrite tools with static
>>> analysis, or just go with aFoo.
>>> _______________________________________________
>>> dev-platform mailing list
>>> dev-platform@lists.mozilla.org
>>> https://lists.mozilla.org/listinfo/dev-platform
>>>
>>
>> +1 for consistency.
>
> At the risk of mostly repeating myself, I want to emphasize once more that:
> (1) We have already made the decision (in other threads) that consistency
> trumps everything else.

If I hear correctly, it sounds like Jeff is arguing that it is not
always helpful to know that a variable is a parameter, and so it
shouldn't be a requirement that parameters are always named to
indicate that they are a parameter.

If I were to always use "firstObject" for the name of the first
object of immediate relevance, then this would be consistent
naming.

If there is only one object, then "firstObject" would still be a
consistent name, but I wouldn't say that using the name "object"
is inconsistent.

When it is helpful to know that a variable is a parameter then
consistency in the convention to identify parameters is helpful.
If it is not necessary to indicate that the variable is a
parameter, then it doesn't matter.

If we choose the conversion to remove existing 'a' prefixes, then
we would need to either automatically move to a different naming
convention that clearly identifies the purposes of the variables
(out params as non-const references, and maybe pointers may be
specially indicated) or manually go through and work out what
names to give the variables that would benefit from clear
identification.  The scope of the manual process of course makes
this impractical.

I think we could relax the 'a' prefix requirement to be a
convention used when identifying the variable as a parameter is
useful.  My opinion is that this is useful for most parameters in
all non-trivial functions, but others disagree.
_______________________________________________
dev-platform mailing list
dev-platform@lists.mozilla.org
https://lists.mozilla.org/listinfo/dev-platform

Reply via email to