kdevelop/languages/cpp

Adam Treat manyoso at yahoo.com
Thu Mar 10 14:27:04 UTC 2005


On Thursday 10 March 2005 7:17 am, Roberto Raggi wrote:
> Hi!

Back at ya! :)

> On Thursday 10 March 2005 12:54, Adam Treat wrote:
> > It wasn't just Roberto's code.  The formatting was messed up all over
>
> Hey Adam, you can read the cvs history/annotate. Most of the code is mine
> :) as most of the code of the catalog, codemodel, parser, c++, java,
> qeditor, abbrev, plugin architecture, the class browser, the quick open,
> buildtools/projectmanager, ... NOW I hope you will not change the coding
> style of these other components without asking to kdevelop-devel.

Roberto, I know you have authored much of the code (thanks BTW!) ...  That 
isn't the issue and never was.

> Anyway, the point: the cppsupport it is *not only* my code and now it is
> *not only* your code. Alexander has contribuited to it, Victor has
> contributed to it, Bernd has contribuited to it, John has contribuited to
> it, Daniel has contribuited to it, and I have wrote it.

Exactly.

> > languages/cpp with several different coding styles all over the place.
> > Some files exhibited multiple coding styles in the very same file.  So,
> > we are left with a situation where the coding style is inconsistent in
> > individual files, not to mention the whole of the cpp language part. 
> > There was no way to tell what the 'correct' style was and no way to tell
> > who was responsible for which coding styles.
>
> you know why? because I don't change the style of the code contribuited by
> other developers. It's something i don't like.

And that is a huge problem.  Why?  Well, because you didn't enforce a 
formatting style across _individual_files_ let alone across a language part.  
This results in unmaintainable code except for the poor blokes willing to 
suffer the migraines and eye bleeds of trying to understand complicated code 
that spans several thousands of lines in some files... all in inconsistent 
format from method to method and line to line!

Had all the code under your maintainence been consistent, I would have honored 
it without a problem.  I couldn't even discern what the 'correct' coding 
style was!  So, we're really we're left with a single question:

"Is it better to have a pile of extremely important, but unmaintainable, 
obfuscated code OR should we demand that a consistent coding style is adhered 
to in an individual file let alone a major module?"

To me, the clear answer is we need to have maintainable code.  Sorry if that 
hurts cvs merges and annotate, but it really is a result of the fact that you 
didn't enforce a consistent coding format on the stuff you maintained.

> I maintained most of Gideon/KDevelop for a lot of time. I'm very proud of
> my code, and even when I replaced the 99% of the old code I always tried to
> use the same *coding style* (and in many cases the Copyright header) from
> the old authors.

And that is great, but it's too bad the contributions you accepted didn't have 
the courtesy to do the same for you.  I mean, if they had, then it would have 
been consistent and we wouldn't have any problems.

> anyway, I really like all these improvements to the code completion and the
> C++ support! Great work guys! I'm very happy you and matt are working on
> cppsupport.

Thanks!

> I'm happy If the new coding style will help you :) but, please use
> kdevelop-devel mailinglist. We are an open community, and there is no needs
> to decide the things on the irc channel or via email.

Ok, fair enough.  Still, I think the outcome that resulted was the only sane 
one.  

Am I really so misguided to think that coding style should be at least 
somewhat consistent from method to method in a file spanning several thousand 
lines?

> ciao robe
>
> _______________________________________________
> KDevelop-devel mailing list
> KDevelop-devel at barney.cs.uni-potsdam.de
> http://barney.cs.uni-potsdam.de/mailman/listinfo/kdevelop-devel




More information about the KDevelop-devel mailing list