Jonas Oreland wrote:
> Konstantin Osipov wrote:
>> * Jonas Oreland <Jonas.Oreland@stripped> [09/03/05 22:36]:
>>
>>> linux kernel has a concept (that I would like that cluster also had),
>>> that I think might also work for mysqld
>>>
>>> they also prefer not to change *all* code in one go due to merging,
>>> history etc.
>>> but they have a perl-script called "check-patch.pl" that verifies
>>> that all new code
>>> that is added is conforming to code standard.
>>>
>>> maybe something to look at...
>>
>> We tried in 2005 to write an indent rule to check that.
>> Unfortunately, our style doesn't allow fully automated checking.
it's hard to believe for such a style to exist. A style difficult to
allow fully automated testing is another thing
but tools that check pretty involved coding styles are known to exist,
not sure about open source, though.
search CPAN?
>> So in order to have such script, we need to be able to change the
>> style. Which brings us back to the question about the procedure
>> for accepting change.
>
> they use a hand-written perl script.
> it can be made to accept any style.
>
> it can likely be made such that it does not give any false-warnings
> but instead miss some real warnings...
>
> there script also works on patches...not files.
> (dont know how they do that though...
I'd run the tool on the original and patched files, then compared tool's
output.
> and it's in perl which looks
> like black magic in my eyes)
bad perl coding style there, likely.
Thanks,
Michael Izioumtchenko
>
> or ?
>
> /Jonas
>
