Core dev meeting #62

Transcription
Transcription

@Howo

All right, so yeah, on my end, I after some discussion in Mattermost, I finally went back and finished the muted reason thing. I saw that bar tag you picked up on it, but I opened two branches, so I closed the one that is bad. So this should be working. I know you said you needed to review it to make sure that Postgres reSql is compliant. So as soon as that's done, it makes it changes as well. Apart from that, meanwhile, I'm still working on the current basis.

@blocktrades

Sorry, I didn't understand. What was the issue about the Postgres compliance?

@brianoflondon

Actually, it was no issue. It was just my question to guys which are implementing switch to Postgres team to verify that new changes made by power, maybe they will need to be somehow also adjusted in new SQL implementations because some parts of Python code were moved to the SQL directly and we need to now figure out if something needs to be extended there. I will consult with them tomorrow in detail and if something could be needed to SQL versions already using the Postgres team version, I will let you know how just to supplement this implementation because it is already much to develop. So I think that would be simpler. Mostly it should be related to extending that will not schema objects by new fields. I suppose that on that part will be potential.

@blocktrades

Okay, so you're talking about changes that Postgres team you're talking about some changes the Postgres team guys have done that hasn't yet developed.

@brianoflondon

Yes, Ofcource

@blocktrades

Okay, so I suppose when they merge, they would probably like.

@brianoflondon

Yes, they will merge, but how it doesn't have them.

@blocktrades

Yeah, yeah, what I'm saying is how it doesn't have their change when they rebase to get how is version when they rebase develop, they'll get everything.

@brianoflondon

Yes, but I'm not sure if just rebase will be sufficient because we can not have some conflicts executed conflict because guys can maybe they added some another file holding similar structures in SQL, which of course was not modified by whole. So we need just review such code and I let him know if anything is needed from his side.

@Howo

Okay. And you think you'll, you'll know tomorrow, right?

@brianoflondon

Yes, I will try tomorrow.

@Howo

I mean, there is no. Yeah, I think that's that's it for me. Also, it's been a while since I worked on how main and the, the speedup has really been noticeable. The, what used to take like an hour and now it takes like 15 minutes. I think that's quite impressive.

@blocktrades

Well, nice.

Interesting. You said, oh, you said you were working on something else too, right? Or maybe I miss it.

@Howo

Oh, yeah, it's, I mean, it's the underlying Rosetta API thing.

@blocktrades

So how's that going?

@Howo

That's not going very well these days. The issue is I'm, I, I'm migrating a way to try to use your, your newest changes. And the, I mean the newest libraries that you presented at Hypefest.

@blocktrades

Okay.

@Howo

The issue is that it's in TypeScript and I initially wrote the thing in Node. And so I basically need to migrate a bunch of the code, onto that. But overall it's going to save me time at the end of the day because, then I don't need to redo most of the operation passing. And because the trickiest part of this entire thing is being able to create offline signings. Well, offline signing while you generate a transaction and then you sign it and you send a sign, a sign transaction, which is something that happens in two parts. And that's not really something that's supported by anything in our tooling. But, I know that your library helps with doing that. So that's the goal. So I'm still in the migrating phase and then afterward I'll be able to, to whatever it is.

@blocktrades

Okay. Do you have any questions about any of the wax stuff you need? Do you need anybody to contact to help? I guess the other thing is I was going to mention to before even getting you answered. I know that Bartek said that they're using like editors, basically, the IDEs that help them use the use wax. And I don't know if you're using if you've got that configured, but that might be useful too.

@Howo

No, I was no worries. I assume it's in Visual Studio?

@blocktrades

Yeah, exactly.

@brianoflondon

Yes. We are using Visual Studio Code on Linux. And as I lately checked, also works there, and developed containers very well. Wax has added one in its configuration and all what you need is opening its folder in given that container. And every tool required to work with wax is in such a container. So you just need to build it and know what you want. So that's really helpful in my opinion.

@Howo

Actually, while I'm at it, how a mature is the Python version of Wax?

@brianoflondon

The Python version of Wax is under development. And we made some significant steps like the first part called BQPy has been finally merged to develop. And right now we are getting ready to start designing and implementing Python object representation in wax to have a similar interface to TypeScript. But unfortunately, such words didn't started yet because most of the Python guys are still involved in other started tasks. But I hope it will be possible soon. So I hope that first version of Python maybe could be provided by to end of the year. It will be not definitely not ready for release, but maybe just show some structural interfaces and some use scenarios we plan to do.

@Howo

Okay, sounds good.

@blocktrades

Okay, so I guess my other question is, would there be helpful for you to have any particular contact for using wax?

@Howo

Yeah, that would be helpful. I mean, it's mostly something that's offline where you work. So if there are on Mattermost or you should literally an email address would work as well.

@blocktrades

Yeah, Mattermost is probably a good place just, I guess, ping Bartek and he can.

@brianoflondon

Yes, I know that I rarely are online there, but we try to be more active on this channel.

@blocktrades

Okay, so I guess that's probably it from your side right now.

@Howo

Yep.

@blocktrades

Okay, I guess I'll take it next. So, you know, obviously, we've been working on a bunch of tools. So in Hive D, I guess one of the things that came up is been desired to increase the expiration time for transactions. So we started looking into that, because right now with transactions basically only allowed to expire after an hour what if they're once they're broadcast? And so we have an idea for how to fix that now. The most obvious way we were a little concerned about at least some of us were, and mainly Andre, but so we've done a bunch of tests to basically test how much congestion we could get by basically generating a huge number of transactions, and seeing what would happen if, you know, if we kept the block size very small like it is now. And then bombarded the network with a lot of transactions to see if extending the expiration time could cause any issues. And it did push up memory considerably. And it also exposed some other issues that basically never really tested for because we just haven't tested that kind of huge traffic amount with small blocks. So, we've made a couple of fixes related to that. And we also have come up with a method, I think is Bartek's idea for how to handle transactions with the longer expiration time but in a way that won't create that kind of congestion. So that's we're going to implement that. And that'll be in the next hard fork. And so as part of all this, since we're making some hard fork-related changes, and still making a fair amount of changes to have debug fixes and such we push the hard fork date back further. So, we've got it set for December 11 at least nominally right now. That's not a hard date, but I think it's a reasonable one that we might be able to make. It really depends a little bit on how much more we decide to try to fit into Hive D and how fast we are able to do that. Let's see what else. There's a bunch of minor improvements to high D but I'm not going to cover all those. Beekeeper is basically Bartek just mentioned beekeeper pie beekeepers basically a new sort of wallet for handling. It just handles keys that's all it's a key storage system I guess keeps an encryption system. And so, as he's mentioned, we've made we've got a Python-based library now for talking to that. We use Beekeeper in both denser the new condenser replacement. And it's also used but the Python versions are being used by Clive because Clive is our Python-based wallet we've been working on for a bit. So that's, that's why that package has been created. And there's been a fair several improvements to the Beekeeper API to support Clive's usages so we learned a bit more about issues with how features Beekeeper needs while we're working on Clive. So I think those are mostly all completed at this point though so I don't know that there's a whole lot going on Beekeeper right now that's new and less Bartek can think of something that haven't that he's still interested in adding to be a keeper. Let's see. So the other thing we've been doing is been making a lot of changes the half. This is for a while now. We made some fixes to the auto detach mechanism the other hatch mechanism is how we handle the cases of a half app dying unexpectedly that used to cause performance problems because data would accumulate. And so we have to basically clean up the app when it dies. And that's what the auto detach mechanism is and we created that a while back but we made we found some problems with this so we fixed them. And we also that that mechanism used to be controlled by Hive D. So, Hive D would periodically basically issue out a statement to see if any of the apps were dead. And we moved that logic out of Hive D and now it's just in a separate crown job. So it's its own thread it just periodically checks in any of the half apps that have died and if so tells have to detach them from the system. We also fixed some a bug we found in half that call could randomly call a problem restarting a half node and that's been reported there's been reports various times of a failure restart half so hopefully this fixed all those issues. And we've also been looking at and improving the process of upgrading half, and that's like the last task I guess half related that I know about before the release. And we're basically getting ready to start that now I guess it'll be completed, maybe a week. Let's see, we've been doing a lot of work in hive mind. It's one of the things as mentioned earlier today earlier in the call was this move towards Postgres T. So right now it uses a Python-based server and we know we get much better performance with Postgres T and it also fits our model for documentation of API etc. So, we're moving hive mind to support Postgres T but in order to do that we have to take a bunch of the Python code that's in there and the API side and replace it with SQL code so that's a big ongoing task. I think it's pretty far along but I don't know exactly how close we are to completion yet. I guess it'll be hard to know until we get really close to really say where that's at. Another thing we did is we reduced the hive mind image size, it was, it was getting kind of big and now it's down to a quite, quite manageable size I feel like it's 20 gigabytes but I can't be sure. So that's the Docker image for downloading I'm talking about, and it had gotten up to be quite high maybe 250 gig or so big reduction there. Another thing we're doing is while I was watching a performance of Half Block Explorer in hive mind I noticed that Half Block Explorer in particular was a lot slower and it turns out the reason was reputation calculations were slowing it down because they run in the same thread. So it is a half-block explorer itself so basically is a purely sequential process. So now we're testing moving the reputation calculation completely outside into its own to its own app. So therefore it runs in a separate thread, and just having half hive mind and half block explorer read the tables, which are basically now asynchronous to the apps themselves. In the case of reputation, it's not really that important that they're exactly matching we just basically need them to be roughly matching. So I'm not caring if we happen to get data from one block before, occasionally in a, you know, every once in a long time. So that's that we're doing that test now and making those changes. And I think we'll be finished. We're basically, I'd say almost finished as of today or tomorrow. And we've already got one test started with the version of the code that's basically got hive mind split out from reputation tracker and we'll do the same thing for block, we'll finish up the work for block explorer and start testing that as well. So we're running a test on that now and it'll probably take about, I guess, two days to get the answers there but I'm expecting no problems really. And that's, those are probably the major things left I would say in terms of the half apps.
I don't know how far we'll get with the Postgres T before release so I'm not 100% sure yet whether we'll be able to release half the new half apps with Postgres T or not so we just sort of sort of waiting to see on that point. I would like to if it's possible, of course. Let's see. It's part of all that highman stuff there's been a lot of refactors and cleanup of the code as well, just naturally when we're replacing the Python code or making a better one. And I guess I mentioned Clive already, we're making a lot of changes to the UI of Clive primarily to make it easier to use. And we've released two versions, which are basically alphas and we're not still to the point where I want to do a big roll out but I think we're getting much closer so I'm guessing in the next month will be will have something that that's pretty good shape for for people to try out. And then beyond that there we're still continuing work on the denser, which is the condenser replacement. And I think that's going okay at this point we're, we finished, we finished most of the work on the primary tool, which you know is the normal tool for looking at posts and everything. And we're just occasionally fixing bugs there. And on the other side the wallet is where we're still doing work, basically the wallet replacement part. And there we're of course getting to test wax as well. So we're testing wax primarily with denser and with half-block explorers UI where we've also been making a lot of changes. And we're testing the Python side with Clive since it's a Python tool. So that's pretty much the state of things right now. I think we're really looking at the December release for everything that's my goal at this point. And we're starting, I'd say production testing of some of this stuff and hopefully this week. So that's pretty much it for me. Any questions about any of that?

@Arcange

Yeah, so you talk about a denser release for testing next week. So there is still no website right now to test it or to see it or it looks.

@blocktrades

Oh, so if you want to test denser the, yeah, well actually Gandalf just posted a link. There's a link for testing it.

@Arcange

Okay, thank you.

@blocktrades

Good question. And for Clive, if you want to test Clive, it's just a Docker image you can download it's really easy there's a post, and you can just go to the B devs post and they've got instructions for how to test it. But like I said, it is still undergoing quite a lot of changes UI-wise and I think we'll have those completed in the next month. One of the interesting things about Clive I'll just mention in passing is, it's going to basically have a lot more flexibility than I think any other wallet has for doing complex operations, especially transactions where you're going to put a bunch of different operations in a single transaction that kind of thing. So it'll be a nice testing tool for someone who's developing and testing won't test interesting experiments I guess to say. But, again, some of that, I think it's probably another month for you to want to play with it seriously in my opinion. Okay, any other questions? If not, anybody else wants to take a chance to speak about what they've been doing.

@mcfarhat

Hey guys, sorry, was a bit late today.

@blocktrades

No, problem.

@mcfarhat

Yeah, so maybe I'll give a few updates since we've been getting more involved with the hive tabs and hive projects. So yeah, mostly we've been doing work on block explorer UI. We've introduced a lot of suggestions at least and some development also from our side. I'm happy with a few ideas that we came through that were already implemented like we, I think you mentioned it in one of the updates, Dan about the witness schedule. This is helpful because it wasn't available as part of the current blog explorer, maybe showing proxy info and user profile. The UI, I think today look this maybe maybe finish this. I think this is very essential, at least for us because active that we are linking to the blog explorer and we host our own instance. So being able to manipulate colors in easy ways so that the UI looks much more integrated and the experience is as seamless would make it much nicer. I did submit a few issues on the dark mode, light mode, and I think they're under work now. Yeah, there were a few. There were, I mean, we're finally done with the value calculations on the wallet card though so that's also fixed. Yeah, I guess. I mean, we're getting, we keep doing some more testing and just to fine-tune things, but overall, I mean, I'm really happy with how Blog Explorer is progressing. We started jumping to Dancer. So, and what was his name? Keiko cot.

@blocktrades

Okay.

@mcfarhat

So, two guys reached out to me, they seemed very friendly. So, we did submit a couple of merge requests. When we started setting up Dancer, we had a few issues, but now I mean, we found out how to solve some of them. We have the environment setup. We still have issues with the authentication module, both the blog and the wallet, they run smoothly, but the authentication module, there is an issue with it. I guess I'll try to chat to the guys, see what we're doing wrong or something that needs to be improvement.

@blocktrades

Yeah, well, if you guys are in trouble or something needs to be improved, either documentation or implementation, I don't know which, but yeah.

@mcfarhat

Yeah, yeah We did submit another one for RTL support, Arabic language, and I think all of the guys assigned a ticket our way. So, I don't know if it's just some markdown testing that needs to be done, or if it's really some coding. I think it's some markdown testing, but I left him a note to see what he really wants.

@blocktrades

Yeah, I saw that too. And I don't know what that is, so I couldn't tell you either, but I did see your question. It seemed like a good question, so we'll wait for him to reply.

@mcfarhat

Yeah.

@blocktrades

I did have one other question. I think, you know, I'd ask you if you had a guy, if your guy could look into the balance tracker calculation stuff, but it sounded like he thought he wasn't the right guy for that.

@mcfarhat

Yeah, yeah, I'm jumping to that. Yeah, exactly. So, yeah, the guy I had who was supposed to be was going to work on the balance tracker couldn't continue with the work. So I only have now two resources who are helping out and neither of them is able to properly and fully set up the balance tracker. So until I have a new resource who can do this, it's pending a little bit for now. If I have some time, I'll try to work on it myself, maybe more, but otherwise, I'll have to find a new resource who can actually help out with this.

@blocktrades

Okay.

@mcfarhat

I mean, like I said, if it's any urgency.

@blocktrades

No, no, it's not. I mean, I'd like to get it going, but mostly I'd like more, you know, outside guys who are capable of doing that kind of work. I'm glad you're looking for a guy like that. I think that's useful. Useful guy to have around.

@mcfarhat

Yeah, excellent. Excellent. So basically, this is it. And I know you mentioned the Clive. I really want to jump into Clive, but we don't have too many resources. I'll see hopefully we have some time to look into it.

@blocktrades

Honestly, Clive is probably not the right time to anyone else. I've already got too many guys working on a cloud right now.

@mcfarhat

Yeah, I got it. And we're already overbooked, I guess, from our side.

@blocktrades

Yeah, yeah.

@mcfarhat

We'll try to just keep working on the blog Explorer on Dancer and hopefully get someone to do some work on Balanced Tracker.

@blocktrades

Sounds great.

@mcfarhat

Okay, thank you.

@blocktrades

Thanks.

@mcfarhat

On the activist side, I mean, we've done a lot of free releases over the last few weeks, and we do put updates there on the swarm of swarms, which is, I think, and that's a good place to put some minor updates, but I'm just sharing about the Hive tabs here.

@blocktrades

Yeah, it makes sense. Okay. Anyone else want to have questions or want to talk about what they're doing?

@Howo

Alright, then I think that's about it. Unless anyone has any questions about any development going on. It's at systems by Brian and Archangel.

@blocktrades

Yeah, Brian maybe had any sound problems of guessing, but.

@Howo

Yeah. Alright, well, thank you all, and see you on Malibus.

@blocktrades

Thanks, you guys.

@mcfarhat

Thank you guys. Bye-bye.

H2
H3
H4
3 columns
2 columns
1 column
Join the conversation now
Logo
Center