Davi Arnaut a écrit, Le 25.09.2010 03:31:
> On 9/24/10 12:22 PM, Mats Kindahl wrote:
>> In this particular case, we would probably be better off if we could
>> turn on -std=c89, which disallows C++ comments in C code.
>> The downside is that it would of course mean that we have to start
>> writing standards-compliant C code (oh horrors!). :)
> I wish that overtime we could focus more on styles and actions, such as
> this, that have more practical consequences. Over time, discussions
> about the cosmetic aspects of coding style tend to get very silly.
Since I'm the one who sent the original proposal, I feel a bit targetted
by the suggestion, so please let me explain.
I have worked in MySQL dev for soon 8 years and have respected the
coding style as requested. But, maybe it's the effect of those 8 years,
I end up tired of some rules, finding them plain unneeded.
Why can I write
int x; // this is an end-of-line comment
// this is a comment alone on its line
why does it have to be
/* this is a comment alone on its line */
I don't know. I pointed out the small problems which it causes. And I
find that learning and keeping this rule in mind is an unneeded burden
put on newly hired developers.
That we should get rid of it in order to simplify our coding style and
free the brain for something else.
When I made my proposal, I expected to receive only "yes of course"
feedback without any opposition, and later have this straightforwardly
approved by the committee and be done with it.
I didn't expect it to launch a discussion. Conclusion: "you never know"...
Same for my "true/false instead of TRUE/FALSE" proposal. I am asking
that we just accept what the C++ standard defines. I think that new devs
would be better off if they didn't have to learn and remember that they
must use TRUE/FALSE. That reviewers would be better off if they didn't
have to watch for use of true/false and reject it.
Again I thought it would get "yes of course" feedback.
I will probably continue making proposals of this kind in the future.
> How about a proposal that new C and C++ code must be
> standards-compliant? That it must compile warning free? etc. Do you have
> an idea of how much work is needed to make MySQL compile with -std=c89?
Why not; it would be great improvements. You are free to make those
proposals, and I see you have already done some real work about that.
I don't see your proposals exclusive with mine.
By the way, allowing true/false or //, is not about being standards
compliant, but close: it would be about being standards _tolerant_.