Re: Cleaning up webrtc spec

> On Sep 23, 2015, at 7:23 AM, Harald Alvestrand <> wrote:
> On 09/23/2015 03:18 PM, Cullen Jennings (fluffy) wrote:
>>> On Sep 23, 2015, at 7:05 AM, Harald Alvestrand <> wrote:
>>> On 09/23/2015 02:57 PM, Cullen Jennings (fluffy) wrote:
>>>> I do not think this is a good idea. It means when you a single word changes in paragraph, often the whole paragraph changes. This means when you go to look at the PR, you have to manually try and figure out what changed in the paragraph. If a person adds a new outer section, the indentation of the every inner section changes and breaks ever existing PR. It also means when the WG is trying to track changes it is much harder for them.
>>> At the moment I can't remember a single case where the WG (or any member
>>> thereof) has said that they cared what the history of a particular line was.
>> I did not say that and you did not read what I wrote. Please do. 
> OK, I have read it again. My response was to the part "It also means
> when the WG is trying to track changes it is much harder to them" - if
> they never tried, it should be an indication that it's not important to
> us whether it's easy or hard.

By track changes I mean understand what has changed to the spec from whatever was the previous version they read. Right now most people are not reviewing the spec start to finish each time it comes out, they are trying to look and see what changes have happened since previous one. Given the state of the HTML diff tools, and that we move sections of text around, it's hard for them to do that. You will note that we don't get a lot of feedback on the list about changes we made. Part of this it is hard to review. What a bunch of people are doing is reviewing the merged PR much like we do when we merge them. 

I can live with whatever you decide on how to do this but here is how I see it 

1) you are at a stage of development of the spec where what you need more than anything is broad review from people in the WG 

2) you are making a change that makes that harder with little to no real benefit. 

Regardless of anything to do with this, we should be encouraging people review the spec. 

>>> The way Martin's suggesting (if I understand it rightly), people would
>>> be encouraged to commit the whitespace changes as an extra commit in the
>>> PR - so if one wishes to dig into history, one can.
>> No - that is not his proposal. 
> My understanding of his proposal is that there would be a presubmit that
> checks for consistency with the result of running tidy, and fails if
> there is a difference (probably with an error message of "please run tidy").
> My understanding of people's response is that they would be likely to
> run tidy, commit, and push the resulting commit onto their PR
> (especially if we told them that was a Good Thing).
> What is your understanding of his proposal?

Martin typically puts them in the same commit as do I because our editors automatically reflow. You recall that is how I used to do it and it caused enough problems for others that I switched editors for this spec so it does not reflow. 

Received on Wednesday, 23 September 2015 13:39:23 UTC