| Weekly edition | Kernel | Security | Distributions | Search |
| Archives | Calendar | Subscribe | Write for LWN | LWN.net FAQ |
It wouldn't be a Weekly Edition without a flame war over style
It wouldn't be a Weekly Edition without a flame war over style
Posted Mar 27, 2008 1:39 UTC (Thu) by pr1268 (subscriber, #24648)Parent article: Quotes of the week
When you reject useful patches based on "this is not our preferred style", you piss people off.
Ahhh, it just wouldn't be a LWN Weekly Edition without more debate over coding style. (Yes, I do realize that this particular one is merely tangential to LWN.)
People often fail to realize that sloppy and/or ill-formatted code has both direct and indirect impact on the reliability and maintainability of a software project. These two attributes of software are incredibly paramount in Linux. I shudder to think of how much time and effort is spent on troubleshooting a kernel panic where the developer has to wade through hundreds of lines of source trying to find matching curly braces, an assignment operator where a == should have been, or (my personal favorite) a stray semicolon at the end of a for( ) or while( ) loop (those are pure evil!).
Get with the program folks. No matter where you go, you will be expected to abide by standards. It's just a fact of life. And announcing that you will continue to willfully "violate the style. Just to spite the fundamentalist movement" is not very professional, nor will it make you many friends.
Here's an idea: Let's all write code in RPG, where each statement (one character each) must go in a particular column and line number without any tolerance for varying styles.
(Log in to post comments)
It wouldn't be a Weekly Edition without a flame war over style
Posted Mar 27, 2008 6:24 UTC (Thu) by lysse (subscriber, #3190) [Link]
> I shudder to think of how much time and effort is spent on troubleshooting a kernel panic where the developer has to wade through hundreds of lines of source trying to find matching curly braces, an assignment operator where a == should have been, or (my personal favorite) a stray semicolon at the end of a for( ) or while( ) loop (those are pure evil!). Yes, I think we can all agree that C is the *very worst* language Linus could have chosen to start coding Linux in - Modula-2 would have been *so* much better. *ducks* Seriously - if a stylistic convention isn't pissing everyone off equally, chances are it's just a reflection of the preferences of the biggest bully in the room. For all of your impassioned ranting that everyone must compromise or suffer in misery and torment, it's not at all unreasonable to expect that one's acceptance into a community will cause some adjustment in its conducts and practices - and nor is it unreasonable to leave that community as soon as one discovers otherwise. If rules exist, they do so as boundaries for discussion and starting points for development, not as decrees fixed for all time. That's just how communities *work*, and programming teams are no different - or at any rate, they *shouldn't* be.
It wouldn't be a Weekly Edition without a flame war over style
Posted Mar 27, 2008 7:36 UTC (Thu) by ncm (subscriber, #165) [Link]
... unless there are millions of lines of code already written to one standard. Choose one: (a) new code matches it, (b) the millions of lines of old code are altered to match a new style, or (c) you get a mishmash of different styles. Nobody will do (b), and nobody wants (c). What's left?
It wouldn't be a Weekly Edition without a flame war over style
Posted Apr 3, 2008 3:40 UTC (Thu) by lysse (subscriber, #3190) [Link]
(c) happens anyway, despite everyone's best intentions. That's community too.
It wouldn't be a Weekly Edition without a flame war over style
Posted Mar 27, 2008 7:48 UTC (Thu) by bangert (subscriber, #28342) [Link]
> to willfully "violate the style. Just to spite the fundamentalist > movement" is equally fundamentalist.
It wouldn't be a Weekly Edition without a flame war over style
Posted Mar 27, 2008 12:15 UTC (Thu) by mingo (subscriber, #31122) [Link]
yes - and in arch/x86 (where this discussion originated from) we do not actually reject patches based on style issues - we either fix it ourselves if it's easy, or we ask people to fix them up and if they refuse it we still do it ourselves. So far out of the more than hundred arch/x86 contributors in v2.6.25 (who authored over 1200 arch/x86 changes since v2.6.24) it happened only once that a cleanup request from us x86 maintainers was refused. In that case we simply cleaned up the patches ourselves. There's not a single arch/x86 patch that has been submitted to lkml that is not in x86.git right now which has been rejected for pure style issues. People actually like consistent code, they like if code "looks nice" throughout a subsystem, and they like the increased maintainability and lower bug rate this brings. It just needs maintainers and tools reminding people of that in a neutral, objective, non-intrusive and non-workflow-impacting but still persistent manner. It's easy to get non-functional components of source code wrong, it happens to oldtimers just as much as newbies. This discussion IMHO is more about those maintainers who have a gut reaction against like the bad news that a "scripts/checkpatch.pl --file */*.c" run brings when they run it over their own files. As someone who has fixed literally thousands of style issues i certainly know that "oh no!" feeling! :-)
It wouldn't be a Weekly Edition without a flame war over style
Posted Mar 27, 2008 12:38 UTC (Thu) by hmh (subscriber, #3838) [Link]
Heh. Shadows of the Debian dpkg mess with Ian. The difference here is that you guys are much faster at the "cleanup the foreign mess and merge" important step, AND that there is absolutely no small hole someone could try to pry open in order to get one of the worst git history mess I have ever seen merged directly into mainline. Kudos to the arch/x86 maintainers.