Can we fork the discussion about style in general into a different thread, so that it doesn't drown out Andrea's proposal?
On Thu, Dec 19, 2013 at 9:28 AM, Gregory Szorc <g...@mozilla.com> wrote: > We have an in-tree Clang plugin in build/clang-plugin that can be used to > deploy more in-depth checking on top of the AST. You can have it emit > compiler warnings or errors. It runs as part of automation, so if you write > a checker, the tree will burn and bad commits will get backed out. A > problem is it only runs on Clang. But something is better than nothing. > There is lots of untapped potential in this plugin. > > Bug 939350 tracks adding a `mach lint` command (really a generic linting > facility). There is a patch in my review queue now. Over time, I see this > evolving into something very useful. Read the comments in the bug. > > My Mercurial extension at [1] automatically lints Python on commit/qref. > It has drastically cut down on the amount of "nit:" comments in code > reviews and thus makes the code review process more efficient and lets the > reviewer focus more on what the code is actually doing rather than how it > looks. As I like to say, "every time I type 'nit' in a code review, Mozilla > is wasting money on my salary." Humans should not be performing work that > machines can do faster and better. > > The scientific community generally disagrees with your opinion that > "checking formatting in general is a massive waste of time." Studies have > shown that consistent code style helps with code readability and thus > maintainability. There is a reason large or mature software projects often > enforce consistent code styles. Keep in mind this consistency benefits > newer contributors more than old due to the knowledge gap. I defer to mhoye > to post links to relevant studies and papers. > > I agree that the state of JS linting is sad. > > [1] https://hg.mozilla.org/users/gszorc_mozilla.com/hgext-gecko-dev > > > On 12/19/13, 8:33 AM, Martin Thomson wrote: > >> If we are talking about checking (a separate thing, in my opinion), then >> we should be talking about linting :) Checking for formatting is probably >> less useful than something more concrete. Having run jshint over some of >> the code, I was horrified at the output it produced. >> >> That said, checking formatting in general is a massive waste of time. If >> we had an automated tool, then I could notice a problem, then say "did you >> run {clang-format,astyle,js-beautify}", but generally not worry about >> formatting. >> >> ----- Original Message ----- >> From: "Till Schneidereit" <t...@tillschneidereit.net> >> To: "Ehsan Akhgari" <ehsan.akhg...@gmail.com> >> Cc: "Mike Hommey" <m...@glandium.org>, dev-platform@lists.mozilla.org, >> "Andrea Marchesini" <amarches...@mozilla.com> >> Sent: Thursday, December 19, 2013 8:11:39 AM >> Subject: Re: style guide proposal >> >> [...] >> >> I think we should increase the checking we do, and make it a habit to run >> the checks before try-servering or putting a patch up for review. It's >> really not a good investment of anyone's time to have to deal with >> trailing >> whitespace during a review. >> >> [1]: >> http://mxr.mozilla.org/mozilla-central/source/js/src/ >> config/check_spidermonkey_style.py >> _______________________________________________ >> dev-platform mailing list >> dev-platform@lists.mozilla.org >> https://lists.mozilla.org/listinfo/dev-platform >> _______________________________________________ >> dev-platform mailing list >> dev-platform@lists.mozilla.org >> https://lists.mozilla.org/listinfo/dev-platform >> >> > _______________________________________________ > dev-platform mailing list > dev-platform@lists.mozilla.org > https://lists.mozilla.org/listinfo/dev-platform > _______________________________________________ dev-platform mailing list dev-platform@lists.mozilla.org https://lists.mozilla.org/listinfo/dev-platform
