Open Telemetry Uncategorized, 12 Oct 2022

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: 2022-10-12 meeting

Description

Open cncf-opentelemetry-meeting-3@cncf.io's Personal Meeting Room

A

A

A

Hey everyone we'll uh give everybody a another minute or two to join.

B

A

um Looks like it might be a pretty late agenda today. Evan, do you want to get us started.

C

Sure so, I'm just looking for input from community members, um but also uh any triager's approvers maintainers on a PR I have out to Mark issues as stale after 60 days of inactivity. Once an issue is marked stale code owners for any components that the issue has a label for are pinged, so hopefully they could take a look and then, after a year of inactivity, it closes the issue.

C

um The goal is to kind of bring some insight as to what issues are being taken, a look at within the contrib repo and then uh so yeah. Basically just take a look at that, um and let me know what you think looking for any input on that the goal is to be uh pretty gradual. Here we can kind of take a look at how this looks and then um iterate on it later. So thank you.

A

I'm I'm curious one question I had was: is it do you know if this isn't The Purge, that any other cigs are taking with their issues or.

C

Not off the top of my head, I would have to take a look.

D

I mean have certainly seen it I, don't know about Hotel specifically, but I've certainly seen it in a bunch of other uh open source repositories and seems like a good way, particularly and actively making something. Stale can be seen as an aggressive move by somebody who who may have written the issue. Whereas a bot coming along and saying nobody's, talked about this and turning it off and they may you know it.

D

It seems to be less it's a way of kind of having a level of remove and still keeping your issue list clean, cleaner, I think it's a great idea.

A

Right is the if there's activity on on the issue after it's been marked still, does this label get removed, I.

C

Believe so I think that's included in the uh the bot functionality. It's the same one that we have for PRS right now,.

A

Okay, and is the idea that they'll just stay open and just label the sale.

C

uh Right now, yes, um the goal is just to take a look and see which issue so during like a regular uh triage session or something like that. Take a look and see which issues are marked as stale and then use that to help determine if a component is um basically not uh being looked at right.

E

Okay, I would definitely put a vote in for closing old stale issues within some time frame, like I think we have almost 800 issues open like if an issue was open two years ago and it's still like actively like debated. You know a couple times. You know in a year like fine whatever, but if an issue like has been opened and no one's commented on it in a year like that, thing's, probably so out of date or like no one's interested in it anymore, like I, don't know, I I would close.

E

Those out I would put a vote in for closing those types of issues out.

F

I would second that I think yeah if nobody's commenting on an issue, even after it's been marked. Still, they probably don't care about it, and if they do care enough, they can either reopen it or open a new one.

C

I would agree. I saw that bargain just said six months. I uh I would agree with that I found as I'm going through old issues in the uh the backlog um they quickly get out of date with the rate of development in the contribute, repo.

G

So yeah, my my thinking is, you said 30 days or 60 days initially.

C

UH 60 right now, 60.

G

60 and then uh from there I think another four months so means in the last six months, no activity at all. It seems reasonable to to end up being close. uh If, as as pointed, if somebody cares in six months will have a day to say something in that issue, just keep it open. If he says, keep it open, you will keep it open. Like that's.

C

A

um Can we can we add a label like close the sale when, when it's automatically closed, I feel like that might be helpful when we're trying to search for things that were closed like if we're looking for a maintained component? If we could filter by things that were closed as stale as opposed to just closed, because maybe they weren't bugs or whatever it might make things a little bit easier? But.

E

The bot puts a comment in there, but we can't search for that, so the label would be nice. It leaves the.

F

Stay label on, though, doesn't it right so can you just look for your issues that are closed and have that label yeah.

G

But but you may close, um you may close still the issues, because you decide as a person that is no longer applied or whatever, and you may not remove the still label from it.

E

That's a good point, though: the issue Creator may not have permission to remove the label, so they may just close their own issue and then the label's stuck on there and we wouldn't. We wouldn't see it.

A

Okay sounds like we have some good action items for moving with this moving forward with this issue. um Morgan, do you want to take up the next item? Yeah.

H

Just really quick, as many people know, the governor's committee election is taking place next week um there was a blog post about this I just pasted a link into the notes for this meeting and I can paste it back here. uh Really just two things to remember: first, is to vote. uh I think the voting opens next week. I, don't know if you can vote yet, uh but secondly is, if you're a contributor to open Telemetry, who maybe doesn't have a huge number of GitHub commits.

H

Maybe you've been contributing to docs and other things like that. Double check that you're actually on the voter roll and if you're not, uh but you have made what you think are significant: contribute contributions to open Telemetry.

H

um There's a process outlined in the voting instructions and blog posts about how to get yourself onto the voter roll, so I'll paste that link now, but just a basic reminder.

A

Okay, thank you. That was the last item on the agenda. Do we have other things we want to talk about.

G

I think Alex I, don't remember, but I know we had a conversation yesterday that you said we want to discuss today.

A

Yeah uh I think we had a little bit of conversation around the idea of splitting uh components into their own modules right.

G

Yeah, if, if otlp should be its own module or we should create a module around the whole exporter directory with the idea in mind that we may have right in the package exporters, some common code between the exporters that we may want to reuse and go mode seems to not be happy. If you have a module like that and inside that, you have other modules.

B

Don't quit have such cases in the country when we have modules module inside module, yeah.

G

We do and that's why that's where I saw that, uh so if you run make uh go tidy, if you run tidy on all the modules, uh you will see the the error that I'm talking about. Okay.

B

So but if you have, if we have shared code like exporter helper or something like that inside exporter, that's fine yeah. You can get make that one module as well yeah, but.

G

You cannot use the the exporter just like uh go open, Telemetry, iOS, slash exporter as a model. That's yeah! It's a Pity, because I feel like it's a good module, so I'm, not sure if we want other option would be maybe to put this into an exporter, slash otlp exporter, so we we put a plural there in the directory. uh I know God doesn't like plurals, but it's in the middle of the path. It's not like the packaging because that's where they say to not use plurals. So maybe that's another option. We put.

G

We move the things into an exporter, so we can save the exporter package. I I shared with you guys, the the link to that error I mean go, allows it because I think you historically had to allow it, but it discourages based on the the error that I'm I'm getting there.

B

Why don't? Why do we need to rename it have an exporter as as a path for the module? Still, okay, I believe why why not I mean why do we want to do you want to rename exporter into exporters plural? If, uh even if it's like just a part of the module, it shouldn't be a problem right.

A

All right, I think the the so the question that I I think what Bogdan was asking so currently I. You know the pr I have here creates an otlp exporter module in in the path exporter, slash otlp exporter, um but that prevents us from then using the exporter as as a module name right like you, wouldn't be able to create a module with just a collector slash exporter, because that that.

B

A

Work anymore, so now the question was: do we move the otlp exporter under collector exporters, slash otlp exporter, so that we can still use collector exporter for things that are common, like maybe the some of the component stuff? That.

F

Yeah I'm confused by this, though, because in in the go SDK we've got, for instance, um slash SDK as a module and slash SDK, slash metrics as a module and both of them have code in them, and we don't seem to have encountered this issue.

G

It's it's possible goal, allows you, but I, don't know if you will allow it for a long time. You can see the error that I linked.

G

We we even have that in the country of Anthony I'm, not saying that it's not possible. I I think it's an anti-pattern for gold.

F

Sure, but I haven't seen this error before either, um which is the thing that's surprising to me. We do this in many places in the go SDK.

G

G

The command there on the internal AWS x-ray package, which has other two other modules inside that's what I get and I, don't think I have any other. You can try, I mean somebody can go and try it in contribute to see. If, if it's just my configuration, that is wrong.

G

And again, I'm not seeing to not I'm just saying I saw this error. Is this something that prevents us or we believe that it we can ignore this error and just move forward. I'm fine with moving forward. I was just trying to have this conversation with Alex, because I saw this and I'm like. Is this gonna cause us any problem? That's it.

B

The shared code that we want to have also can be as a modules inside exporter path. Right now we have exported helper right. We can keep that one and we put exporter, slash exporter path there, but no, no! That's.

G

Ideal, it's a good name that I would like to to have one option to say exporter that that's something like I I feel like. We are talking about exporter. We are talking about this and we cannot use that package. It's a Pity again I'm not suggesting to use it right now, but I don't want to block me to do that in the future.

A

This this will also be the same pattern for all the other components that we're that are planning on splitting right. So it's not just the exporter, but.

G

Correct correct, correct yeah: we will do the same thing for receiver for processor, for all the things extension as well right.

A

So I guess the real question is: is this warning that you're getting when Com when running tidy, something that we need to worry about or something we can just ignore? Exactly and I, just I just ran it locally and I get the same in that same module, so it it's not just you.

F

Know I I get it as well, but I don't get it in in other places where we have the same pattern. So I'm not sure that this is saying it may be saying something other than what we think. It's saying. um That's.

G

Fine, if we investigate Anthony- and we figure out that it's it's a different problem and we don't care about these I'm perfectly fine to move forward. Just for me, I want to make be seen and make sure we don't. We don't block ourselves here.

F

Yeah sure I'll go ahead and put some information on this ticket about what we do in the go, SDK and uh places where we have this, where we're not seeing this error. Hopefully that will provide some information as well.

A

Yeah and I'll spend a little bit more time. I I know you suggested that I look into it. I'll I'll look into it today, I started, but looking at what I would take this pled, all the components in the exporter packaged into their own module um and hopefully I can I can get some answers here.

G

Would be good, otherwise, an alternative if this is a real problem, an alternative uh would be to use exporters uh for for all these packages, for the all the exporters in the middle of the path like as a directory. But again it's an alternative. Let's stick with this, if possible, and not a problem, Brian you're, you're, trying to say something.

I

Yeah we're saying the internal AWS x-ray module and contribute right that we see this here. Correct I, don't see the air when I do it.

B

Same here actually I don't see anything. What.

G

What version of gold do you have? I have lightest 190.

G

Because I started to see it.

I

I'm on 118.7 I.

G

Think I could do that I'm great to 119.. That's I think when I start seeing this it was a surprise I'll. Do that all right.

G

We can have this debate in the issue. Again, it's not I, don't I, don't feel strong about this. I just want to make sure that we are not doing a stupid mistake here, like.

B

A

Yeah that sounds good I think we have some some steps to move forward.

A

Any other topics- I, don't know if anybody has added anything to the agenda, doesn't look like it. No.

G

I think we have a release today. I, don't remember who is on.

A

Charge for that yeah tigrant Tigres in charge of the release today, okay,.

G

I will talk to him.

A

Sounds good thanks.

B

A

B

Thanks foreign.