Rust Programming Language Rust Cologne, 6 Jun 2016

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Rust Meetup Cologne (2016-06-06): Andre Bogus - Clippy: Survey Results

Description

Rust Cologen User Group - Meetup June 2016

More on http://rust.cologne/2016/06/06/rust-anniversary-part-2.html

Help us caption & translate this video!

http://amara.org/v/2Fi8/

A

Okay, everybody yeah, usually I'm talking about something I, do answering questions today, I'm going to tell you what you folks told me, because we did a survey as is customary with rust projects, it seems and yeah. So we also did a survey first for those who do not know what Clippy is it's a project to create a number of Flint's for the rust language?

A

For those who don't know what lint is it's basically a piece of static analysis, software that will look for patterns in your code that are, for whatever reason, problematic and, as you can see, we have a good number of them. So the current number is 152 yeah. So we check a lot of stuff and.

A

So, let's see what people answered first of those who answered which are as of now 95 responses came in and about two-thirds of them actually use Clippy or have youth club in the past of those two thirds it's around 50 50 either use cargo Clippy or an dependency optional dependency in their project.

A

A

So the question was what is difference between cargo Clippy and an optional dependency? Cargo is, as most of you know, the package manager for rust and it supports subcommands. So you can install a binary into a path that is then called by cargo on the project and we have created such a sub command. You can install if you install Clippy.

A

As of now, you will be able to just type cargo Clippy on your console and then it will run Clippy on your project and you can also use clip es in the dependency of your project and then or as an optional dependency, which you can activate with a feature flag that you will have to put in at the the cargo startup and yeah. That's basically the two options of using it. There are also some more complicated incantations you can use to run.

A

The clip is somewhere on some code, but that's basically the the two ways of running it. So either you just tell it run it. Don't matter, don't worry about the code or you have it in the code and it gets run anyway or by feature okay.

A

So then we asked about what glints are most problematic, so the most problematic Lind at that time was new without default, which I can tell you is now fixed, hey I'm, happy about that, and it's a lot of fixing myself. One of the problems was that clip II asked people to implement the default trade for types they only use internally and those default implementations don't give them any value, so I restricted it to public items. Also, there was one who said well, my new method has side effects.

A

If it was any of you, please rename your method. Please yeah. Also some people have problems with cyclomatic complexity. Well, that's a word but I, always given not in my tongue, which basically wants about methods that get too large, too big, too complex to navigate and well. Some people have problems with that. Apparently they're writing very complex methods. Yeah we're not fixing that if you have problems that you can simply allow it.

A

Then we asked about favorites, there was no clear winner: everyone has their beloved lint or their most hated length. I, don't know they didn't tell us that, but I guess, since two thirds use Clippy we should be okay.

A

Then I asked what is your biggest wish for clipping and the biggest wish, but most people was quick, fixed functionality or a functionality to apply the suggestions but could be directly to their code, which is to say least, a bit complicated, but yeah we're working on it. I also started a discussion with John Turner over Twitter and stack overflow and ISE and yeah basically we're on it.

A

Also people asked about Clippy unstable, which obviously won't fly this year, because this is a compiler plug-in which uses the high-level internal representation hir, which is yeah. Apart from mir the medium level intermediate representation, the most unstable thing you can think of so I guess I'm not sure if we will see a stabilized Clippy before the end of the decade.

A

Anyway, there is now rest up, which is a great tool. You can use to have multiple rushed two trains. You just go to rust up dot, RS and follow the instructions, and then you get an option of installing an idea: Clippy, stable Clippy and whatever Clippy you want not to be part, rust, plus cargo 2, Chainz site-by-site. You also can manage cross-compile us with that and stuff. It's really awesome, and if you don't have it, you should grab it now. There's an open Wi-Fi here you can well ok! So that's about rest up and yeah!

A

Then we had a question number 6. There is no question 6 and yeah. It was simple, it was you just have to click, got it and then there's some people and it's pretty selected by the way. So it's really hard to not answer it, but some people did anyway.

A

I have no idea how or why but yeah that's how it is, and that's basically everything people told us so this gets is a really short talk and thanks to everyone who answered the survey and I also want to make some advertisement will have a vine mind rust, regular table june, seventeenth at the textile frankfurt. It's a really nice restaurant. So if anyone is in the frankfurt area during that time, we'll be happy to meet you there, yeah there's a question.

B

A

Still open, but I think I'll close it soon.

A

Any other questions.

B

A

So I guess around a third of those 195 issues are suggestions to improve existing linds or real bugs that will have to face like for positives or false negatives or bad suggestions. Stuff like that and the rest are linds. We haven't come around to write yet so yeah I think about 120 hundred thirty blintz we haven't written yet, and these only the things we have thought out about. For now. The search space for possible ins is yeah pretty much out there.

B

A

B

Seems just from the end that it's pretty difficult to write and more like?

B

Are you guys planning on making the platform to integrate once more accessible so that people can write their own sort.

A

Of checks for projects.

A

The question was: if we could make it easier to write, custom links and actually I have thought about. That's we. We have I, think one or two issues even on the issue tracker, because there are projects for other languages, I'm thinking about PMD, for example, which user something like a structured description. Language fall in that then gets compiled into those yeah methods to actually do the Linton, and we are thinking about that. But it's since the interface we're working against is quite unstable.

A

We have so much to do just keeping up the current lens and writing a few new ones that will get us quick gains. So I think this is a medium to long term project.

B

Because I miss the beating- that's not the intention of the project that every can write. Little rule checkers for their.

A

Own project: well, you can it's! Okay, since you've missed it and the first rust code, I've wrote was the equal operands lint for Clippy. So it's not too hard. It was the first piece of rust corrode and it wasn't. It was really easy. So if it's a bit wordy the interface you'll have to wrap your head around this, this syntax trees, but once you've done that it's a piece of cake basically.

B

An extension to the existing Vince and the rest Empire, which already has a couple things, so you basically everything else that might be too specific for the western part itself, basically I said, probably have a place in cookie yeah.

A

Also, the good thing is that we can develop Clippy apart from rust, see so we don't get those half an hour compile times which is really speeding up the project.

A

So we have to think about two or three minutes now, if you, if we do all tests, so that's still workable and if you do a specific test, it's even faster. It's a matter of few seconds, so it's much more about turnaround times then yeah having it in NYC and the other thing is, if we make it more complex by having some pre compilation, steps or so also, development will get slower, because the some code has to compile this stuff and yeah.

A

This will increase our turnaround time actually.

A

Correctly, kappa deuce plan to split those.

B

Up into smaller groups.

A

This is a bit difficult for me to decide because there are so many different groupings we could have, for example, okay. Those are probably readability lens. Those are performance lens. Those are security or misleading code lens or whatever.

A

But thing is those categories are not as clear-cut as you would make it out to be so yeah I think if someone wants to come around and tell us a possible categorization will be all for it and discuss it, but so far most of us have been more happy with writing mullins and then too yeah put them in separate drawers.

A

Any other questions. Okay, then, thank you, folks,.