Open Telemetry Uncategorized, 14 Sep 2022

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: 2022-09-14 meeting

Description

OpenTelemetry PHP SIG Meeting

A

A

Hey you can hear me.

A

Good day, everyone wait until 805 to get started.

A

Hello, everyone, hey everyone.

B

Hello there, everybody.

A

B

A

A

All right, it looks like we have built a forum good morning afternoon evening, everyone, let's get rolling through our agenda, for today I share my screen so that you can let people see it.

A

Can everybody hear me? Okay,.

B

A

Hey hello: anybody here.

B

I can hear.

A

Yes, we can hear you.

C

Everyone hear me now.

A

C

A

C

Hear you, oh that's very strange, can you all see my screen? Yes,.

A

Yes, excellent.

C

Hello, everyone knocking here everybody thanks, sorry for technical difficulty. I don't know what happened all right: open, pull requests.

A

C

So we have one from brett seven hours ago right, you wanna talk through this.

B

uh Yeah just quickly there was a there was an issue for um that was blocked awaiting the latest um protobuf um libraries. I've had those for about a week now so um yeah just implementing it. To do that that you know left behind uh some weeks ago. Okay,.

C

All right, so this just needs a review on the merge.

B

Yep, it's pretty short.

C

I'll pin it and do that after this meeting um device it looks like you, you have a couple: um do you want uh looks like you and brett were talking about um backwards, compatibility for this abstract fan to spam.

D

um Do we need this if we go into beta, do we want to start with some old stuff that we don't want to spread in the future, or should we have a clean start with the beta?

D

So I'm not too sure. If we need this, you see compatibility right here.

B

I I only picked that one out just because I I know that abstract spans pretty commonly used, um so it's really just as a courtesy but okay, really so that everyone's code doesn't doesn't break. um That said, I I do think we could go back and drop a lot of our old pc stuff. That's been there for c, more more than one release.

C

What do you think, tobias you um do? You have strong feelings.

A

Not that connected okay. Thank you very much.

C

See it looks like you add this. Had this add alternative information, abstraction.

D

Yeah, that's global access to dresser provider, meta provider because, with the current instrumentation approach, I'm missing a way to get some new instrumentation.

C

Okay, is this ready for review.

D

It can be requested. Okay,.

C

I'll review it today, thank you.

C

Using array lookup instead of recursion in context, oh man, I loved writing that con that recurring part, but I guess it was um as a potential leaky memory situation. Thank you for fixing that right. Would you like to merge this.

B

uh So I reviewed it and and think it's okay to merge, I'm just waiting for uh nice to say. Yes, it's good to go.

A

Yes, it's good to go all right.

C

We can merge that in.

B

Yep, I I never know whether there's more commits or or not coming, yeah.

C

A

C

B

To buy so um yeah, it's probably worth saying: I'm done well, it's yeah! Something like that, just so that I know, because I'm there's there's a few like that that I'll think I'd be happy to merge right now. uh If you're finished, developing.

C

Do you all have a do? Y'all have a normal social contract for that at work. uh This is just an interesting topic to me in general,.

C

That's the only thing I don't like about asking for reviewers automatically is when you get to the point where I understand why they would do it for a public repository, because people don't know who to ask and the code owners thing is really nice. But it is also nice to be able to ask for reviews using their viewers, bitten piece.

C

Don't hesitate to ask for reviews ever I'm sure, brett and myself and others would be very happy to review all the time.

A

C

There is this one for adding interfaces for context and context to uh context key sorry, uh tobias. It looks like this. One is almost ready. I I think it might actually be ready. This one is.

D

Three, it's ready, but it will have conflicts with that up here. So yep got a fix out of them.

C

uh This one's an intermediate transport layer between the http, grpc cloud and exporter looks like uh right. It looks like brett approved this. Is that another one that we think that we're ready to merge.

C

Goodbye slash brett.

B

uh Again, happy to happy to merge it now, um um if devices finish developing.

D

Yeah, it should be finished. uh We should do some cleanup of the exporters afterwards combine http and grpc otlp exporter, because they can use the same code now and just different underlying transport yeah. This can happen afterwards.

B

Right another one for your merge. First.

C

Merge fest all right uh span contacts this one's ambers.

E

E

Hello yeah, actually, uh one of the requirement of the ticket was to uh create a new factory for uh factory methods, so dedicated factory, so yeah, so tobias is suggesting that we should not create a new factory class. Basically so means basically I'm confused whether we have to create a new factory with the class or once what should the final change that should like. I should do.

C

So you you're talking about the sport point though email put in I mean it's not. I don't know if that that's absolutely necessary. No, that might have just been a suggestion uh device. You have strong feelings on that.

D

um Somewhat especially with the new prm I created, which adds consistency with spam package and core, I don't think we should go the opposite way with spam contacts then, and have the factory methods etc for every interface in the class, without interface, suffix.

D

So I would prefer to keep them in the span context class.

C

Can you work with that amber.

E

Okay, then I'll make the changes. I will change this and update. Oh thank you.

A

C

Appreciate that thank sexy, yeah.

A

uh Yeah, so in this ticket, actually I have implemented the retry mechanism in the spanx.

A

I guess in the last. In the last comment, it was suggested that I I should move the logic retry mechanism from trade to the actual exporter.

A

If you go a little bit up.

C

Yeah, it looks like you said: you're gonna, move the logic to the exporter.

A

Yeah yeah, so actually, unfortunately, I got busy with some other priority here. So but uh this economist.

C

That's quite all right. We um we aren't paying you to work on this, so we're very.

A

Happy to get your.

C

Contributions when you're able to give them thank you.

C

The client auto instrumentation for contrib. I know that we're working in that separate channel for the auto instrumentation does anybody have anything they want to talk about with auto instrumentation bits and pieces.

B

uh The only one for me is that um uh have a pr in to have a new repository created to hold the extension. um Everyone seems happy that that extension is, is the the way to go. um So once a repository is created, um uh device and I'll work through how to get the code into the repository.

B

uh And then I think we've got a little bit of documentation, um for you know how to install it, but then we we can really get moving on creating some auto instrumentations.

C

Wonderful thanks for the update brett. That's I'm really glad that we're working on that that's a very commonly requested feature and I'm glad we're moving towards it. I can't thank you enough to bias you well for all of the work that you've been doing lately. You've been working so hard on this project and we really appreciate it, but that's one that really stands out. Thank you.

C

All right, um I guess that covered these bullet points in the auto instrumentation bullet, uh thanks for the update brett.

C

So I guess the last thing that's on the agenda for today is uh talking about blockers for the beta release. It looks like people have added us additional bullet points on things that we should theoretically talk about.

C

I will open up the conversation with supporting for php 7.4 as it's going end of life for security support on 11 28 2022..

C

I can also. I would also like to comment on that. uh I have a vested interest in keeping uh support for php 74 for a long time, because that's the version of php that my company uses across the board- and I don't see us going uh getting updated past 7.4 before 11 28 2022.

C

So um what what do others think? Is there a reason we should just mix php 74 support, just because it's going end of life.

A

At least for beta, we can keep it because we'll be going beta soon right. This is not till two months right.

B

Yeah, I I my only argument for dropping. It would be that you know we can improve our code base um a little bit. You know, drop some php, seven, more isms um and yeah. That's probably. I know that there's a couple of like ffi and fibers um don't work in seven four, but we have workarounds for those.

B

So now I don't have a strong opinion, except for a reduced support matrix because 8.2 will be coming in at around the same time. So that's an extra um extra version to to worry about.

C

Got it okay, um yeah? I think that maybe I think maybe we should keep it for now and then, when we maybe we should remind ourselves at the end of the year, to sort of discuss the php landscape and see how how we're feeling about removing it. Does that work for you, brett.

B

Seems quite reasonable, I mean to do. Nothing is to continue to have seven core support. It works now. It's it's the main sort of version that we develop against so.

C

I I agree with you. However: um it's always you know it's always additional. Like you said it's always additional toil to at to maintain something that doesn't doesn't need to be maintained or that that is bit riding, but I don't think we're to the point of bit rot with hp 74. Yet.

C

All right metric, sdk behavior with 7.4 and 8.6.

C

uh Who added this to the agenda? Did somebody want to talk through this.

B

That would have to be you, wouldn't it tobias talking about metrics.

D

Yeah, um I had to add in work around 47.4, because meatmap was added in 8.0 and there's a underlying bug in php with destructors and weak references, which means that reported metrics might differ slightly between these two versions or how long we report metrics that are no longer active.

D

It's not a big deal, but it's a slight.

B

Difference sounds like it could be managed with a small amount of documentation for a warning somewhere.

B

C

We have two api related issues: 723 and 795.

A

uh Simplifies locking and allows using enormous contacts key.

C

Tobias, do you feel like this is something that you would like for the bay to release these two things? Yes,.

D

But the piazza already and shouldn't delay the beta.

A

C

C

All right and then the last is remove default values from text map, propagator interface,.

D

The current uh implementation was written with a ways in mind and not with requested destructions and co. I think we should put the at least the default um propagation getting setter and make them required arguments which would allow us to add strong typing or generic typing, with sign to these methods.

C

That sounds great. Do you have a did you create an issue for this, or is this something that we need to create an issue for? No, I don't know yet okay, but this is something you believe that we should do before our beta release. Yes,.

D

Because it will be a breaking change.

C

And let's take it up and try and get it done before the beta release. That sounds good to me. I think that anything that can help um anything that we can do to not have breaking changes after the video release will be a good thing.

B

Agreed, let's get out there and break things now.

C

Let's get out there and break that's that's.

B

A

C

I like that attitude brad, that's that's the right attitude, all right. Those are all the things that are on the agenda. Does anybody else have uh non-written agenda topics that they'd like to bring up today.

A

Going once going twice.

C

All right well, we'll see everybody next week, then.

B

Thanks for coming, everybody see you next week.

A

Bye, everyone have a good.

B

A