Notification service №1

, let’s create spreadsheet with certain features and ticks with yes/no

Yes, I will do it, and let’s make the discussion of projects as open as possible

By performance test - 100% in top will be all solutions on languages with compilers…

Not really. Stress testing is not only about performance, but also about system reliability. Will there be OOM? Will messages be lost on sudden reboot? etc.

I am very upset that the jury did not vote :confused:. A lot of work was done by all the participants, but we did not see the votes, even after 6 hours of defense in the Zoom conference .

1 Like

I really like the approach and am looking forward for results!

Who is responsible for this meeting? Since when did the judges stop voting for contest entries? Did you adopt the practice from the design sobgaw? Since when does our blockchain no longer need programmers? What is this mess with contests?

1 Like

Last week was discussion about prolongation of voting for 2 weeks. When we can expect it?

We will have another discussion in two weeks about rescheduling for another two weeks

Revote contest proposal

Introduction

After the end of the contest, we can see that out of 12 jury members, only one voted for the works. It seems that this is not enough for an objective assessment.
From other side, recently, a new jury contest finished so we have several new jury members to evaluate the works.

Since the original jury of the contest did not serve their duty to take part in AMA session and to evaluate the submitted materials, I propose the next.

Proposal

Assign the following jury for the contest and set the end date for voting on November 30, 2021:
@anovi
@hi_artem
@nerzh
@SergeyEgorovSPb
@vadimrusakov
@fabrice_dune
@Ovenstent

If any of them do not have the opportunity to devote this week to study ALL works in this technically challenging contest, they will have to vote “abstain” for all submissions.

After this vote, participants will receive awards in accordance with its results.

4 Likes

Great news!

And dear judges: ask – don’t assume!

отлично! пора заканчивать

All participants, please check that your software are currently running.

If someone tested the tonactions provider yesterday, repeat the test. There was a misconfiguration on my (TON-LABS) side.

1 Like

Did you restart kafka? I mean that connection was lost in this time or not?

I don’t understand your question … I think everything works (at least, monitoring says so).

I have asked because we had checked the own service also, all is working. I want to know what happened on the external service to know that if Kafka will be down, our service won’t work, but it will re-launch in the auto mode as soon as a connection will be restored.

Hi all.
Someone suggested:

, let’s create spreadsheet with certain features and ticks with yes/no

I read all the works. Theу are very different, from sketches to finished projects, but both formally implement all the requirements, that’s why it is almost useless to evaluate the work simply by marking the implemented features as “yes / no”.

Here I have described how I will evaluate projects: Evaluation_criteria.pdf - Google Drive
If I missed something, let me know Telegram: Contact @hi_artem

1 Like

@Ovenstent , thanks for reviewing my submission!
I’d like to claim, though, that your statement about “single subscription request” isn’t accurate in case you mean that one user cannot have notifications sent to multiple URLs (i.e. it can, in fact).
If my assumption is incorrect, please let me know.

Want to answer to the some judge on comment - sqlite, no mention how to scale this app .

Just run several instances and set up a load balancer between them. There doesn’t need to set up and manage any external DB. Very quick and robust scaling. This calls “horizontal” scaling strategy, in opposition to “vertical” strategy, when you need to support many related DB. “Horizontal” scaling doesn’t have bottlenecks apart from network issues, you manage only “autonomous agents”. As soon in architecture appears DB cluster with “master-slave” logic it is problem for a big scale.

I would also like to mention that part of my work, namely Telegram bot for testing, has been neglected by all the jurors but one. Nobody tried to test the public setup in any way, except one juror who gave a try to both public and local deployments.

This was a time consuming and technically demanding contest: having my work partially ignored as a result is demotivating and makes me a bit more confident that something isn’t right about contests, judging process and/or SG in general.

Hi! it was my comment about scaling

Horizontal scaling, sharding, sharing nothing, etc - are very good concepts, I am totally up for them, but they are not for free.
I can’t start one Raspberry Pi, and when the load spikes, just start another 100, and they will magically complete all the tasks.

I believe it is possible to create a performant and robust architecture with sqlite, BUT doing it from scratch can be very challenging, needs code and testing.

This is one of the reasons people choose solutions that can scale out of the box.

Don’t need to scale sqlite - it is just storage. You can run our application on Raspberry Pi , because 7mb memory in idle mode. You can run 1000 instances in the same manner and all will work… Please, show us solutions that can scale out of the box. 100%, that needs to set up additional databases and “write a code and tests” also. I guess it is all about clusters on Mysql, etc… Our solution just needs to deploy many times. I won’t argue more, because we have discussed already and you mentioned that prefer only old “tested” libraries, even if it uses “old” approaches. Thank you for your work and time.