[Webkit-unassigned] [Bug 197973] 'analysis-task-configurator-pane' does not update when switch from one analysis task to another

bugzilla-daemon at webkit.org bugzilla-daemon at webkit.org
Tue Jun 25 02:04:03 PDT 2019


--- Comment #20 from Ryosuke Niwa <rniwa at webkit.org> ---
(In reply to Alexey Proskuryakov from comment #19)
> What is the convention for === in Web Inspector? We should certainly
> consider consistency within the WebKit code base.

https://trac.webkit.org/wiki/WebInspectorCodingStyleGuide doesn't mention this although they seem to mostly use ===. However, I'm not certain that's relevant given there is basically nobody who works on both perf dashboard & web inspector codebase. Quite possibly I'm the only person who ever contributed to both. I do see an argument that using the uniform style across JS code in WebKit is beneficial but that kind of discussion & deployment should happen in a completely different bug than this.

> > That's precisely the kind of data I don't care about. I really don't care if someone with Ph.D. in computer sciences or someone coding in JS for 30 years would tell me what. The only thing I care about whether something makes sense for this particular code base, and I don't think it does.
> This is surprising. What about this code base makes it different in this regard?

Because it's a different codebase than anything else. Any style or paradigm of coding should be judged based on its applicability to each codebase.

Consider goto statement for example. Used incorrectly, it could create a bad spaghetti code. Used correctly, it's a very effective control flow statement. Similarly, in WebKit, we use C++ style type cast but there is nothing wrong with C style cast if deployed properly. C++ exceptions are never used in WebKit codebase but it's totally possible to write a good C++ program with them.

In general, I find these kinds of "best practice" preaching to be extremely misguided and harmful. Dogmatic following of such best practices seem to often result in bad code. Take design patterns for example. Many people who deployed design patterns in WebKit before Blink fork ended up making WebKit's codebase a lot worse by introducing many layers of unnecessary abstractions. People who deployed "const" correctness in WebCore's Node code seem to have only increased the code complexity for a little to no benefit as far as I can tell.

You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.webkit.org/pipermail/webkit-unassigned/attachments/20190625/8b250fd0/attachment-0001.html>

More information about the webkit-unassigned mailing list