[llvm-dev] RFC: changing variable naming rules in LLVM codebase & git-blame

[JF Bastien via llvm-dev]

> On Jul 28, 2019, at 4:08 PM, Chris Lattner <clattner at nondot.org> wrote:
> 
> On Jul 23, 2019, at 9:17 AM, JF Bastien via llvm-dev <llvm-dev at lists.llvm.org <mailto:llvm-dev at lists.llvm.org>> wrote:
>>> On Jul 23, 2019, at 8:30 AM, James Y Knight via llvm-dev <llvm-dev at lists.llvm.org> wrote:
>>> 
>>> As a very frequent explorer of history, I really don't think this is
>>> as big an issue as it may seem. Even absent refactorings, you often
>>> run into the "wrong" commit when looking at blame (e.g., someone just
>>> added a comma rather than actually changing the code you care about),
>>> and have to look past that, to another previous commit.
>>> 
>>> Any interactive blame tool ought to have an easy way to do this. For
>>> example, in emacs's annotation mode (which is what I use), you just
>>> press 'a' with the cursor on the line in question to re-annotate at
>>> the commit previous to that.
>> 
>> You might very well be right, but your answer sounds like you *think* it’ll work out. For a huge change like this I’d like to *know* for a fact that you’re right. Changing variable naming will touch every single line except comments, so it’s quite different from what we usually see today.
>> 
>> Artem mentions a new git feature that’ll do this automatically. Again: sounds great, do we *know* that it’ll work?
> 
> Hi JF,
> 
> As discussed on other posts, there is good reason to believe that the forthcoming git feature will cover this case.

Hi Chris,

I’m happy there’s good reason to believe, but I’d like to know it’ll work. We’re not in a rush, and it’s something I’d expect folks involved in moving this proposal forward to demonstrate. Ideally it’s something that LLVM’s .git file could even default `git blame` to: ignore the giant rewrite commit by default.

However, this is a small concern I’m trying to express… let’s ignore it for now and focus on the rest:


> We aren’t the only project that has been held up by similar concerns about sweeping changes adversely affecting history spelunking.  Similarly, as James Knight points out, this isn’t really that big of a problem in practice even without the feature.
> 
> Sub projects in LLVM have gone through similar phases, including LLDB which switched from 4 space indentation to 2 (as well as 80 columns iirc), which affected literally everything.  This was unpopular, but was a good thing to do, because commonality and consistency across the project is important.  I’m not aware of a show-stopper problem even this massive of a change caused.
> 
> I think that Rui rolled this out in an incredibly great way with LLD, incorporating a lot of community feedback and discussion, and (as you say) this thread has accumulated many posts and a lot of discussion, so I don’t see the concern about lack of communication.

I think there’s lack of proper communication for this effort. The RFC is all about variable naming, with 100+ responses. Sounds like a bikeshed I’ve happily ignored, and I know many others have. Even if you don’t think I’m right, I’d appreciate a separate RFC with details of what’s actually being proposed. Off the top of my head I’d expect at least these questions answered:

What’s the final naming convention?
Will we have tools to auto-flag code that doesn’t follow it, and can auto-fix it?
Will we clang-format everything while we’re at it?
Will we run clang modernizer to move code to C++11 / C++14 idioms while we’re doing all this?
What’s the timeline for this change?
Is it just a single huge commit?
After the monorepo and GitHub move?
Is there a dev meeting roundtable scheduled?
What tooling exists to ease transition?
Out-of-tree LLVM backends are a normal thing. They use internal LLVM APIs that should all be auto-updatable, has this been tried?
Some folks have significant non-upstream code. Have they signed up to remedy that situation before the deadline (either by upstreaming or trying out auto-update scripts)?

LLD and LLDB are indeed good small-scale experiments. However, I think the rest of the project is quite different in the impact such a change would have. LLVM and clang expose many more C++ APIs, and have many more out-of-tree changes (either on top of upstream, or in sub-folders such as backends or clang tools). They also have many more contributors affected, and not all those contributors have the same constraints, making this much more complex. So far this discussion hasn’t seemed to care about these concerns, and I’m worried we’re about to burn a bunch of bridges. Maybe I missed this part of the discussion in the 100+ emails! Sorry if I did… but again, a simple updated RFC would solve everything.


> The reason that I personally pivoted to believe that a “do everything all at once in one commit generated by Rui’s script” approach was the right thing was seeing how lld contributors were able to cleanly move forward even when they have significant out of tree changes - they merged to the N-1 commit, then ran Rui's script on *their* trees, and were caught up with commit N.  This seems to have provided a very smooth transition path, where the conversion was merely a bump in the road, not a major mountain to climb over.

That sounds fine to me. I just want to make sure everyone knows they’re about to have to climb a little hill. Surprise hills are the worst.


> In any case, I hope that we as a community don’t end up in a place where we are afraid of doing sweeping changes that make the codebase better.

Sure the current naming is ugly and inconsistent, but honestly I don’t really care about naming. The one upside I see from this change is the same clang-format has brought: we can finally stop picking nits and arguing about stuff that doesn’t matter.

I agree that we should be able to make sweeping changes. I’m saying that I don’t think the direction I see us going in right now seems like it’ll work out well, and it seems easy to address my worries.


> It is important to continually reinvest in the health of the codebase, and it isn’t bad to incentivize people to merge their changes upstream.

Agreed. However, as you know very well, merging upstream changes isn’t something that’s done over a weekend. First people have to know that a large change is coming and they probably should plan for upstreaming. Second we should give these folks enough time to do the upstreaming, and leave them enough slack in case they fall behind.


> Being overwhelmingly afraid of ‘breaking history’ is a path that leads stagnation and ultimate replacement by newer technologies.  There are many other projects that have suffered this fate, usually for a combination of reasons including things like this.

Again, this concern is small compared to the others I’ve expressed. Let’s have it as one item on the list of things we want answers to before moving forward. I think the rest of what I’m raising is much more important.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20190729/58450e3c/attachment.html>